mirror of
https://github.com/Imbad0202/academic-research-skills.git
synced 2026-09-14 13:51:17 +08:00
main
31 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Edward Cheng-I Wu
|
f1a57bbcab |
fix: shared file-lock helper with msvcrt backend for the remaining fcntl sites (#845) (#847)
* fix: shared file-lock helper with msvcrt backend for the six fcntl sites (#845) scripts/file_lock.py owns the backend choice (fcntl.flock on POSIX, msvcrt.locking on byte 0 on Windows) and routes adjudication_activity, inquiry_branch_ledger, review_criteria_binding, and ars_mark_read through acquire()/release(). POSIX lock sequences are unchanged. Per-site Windows decisions: adjudication reads degrade to exclusive with a 5 s bounded wait; the review-criteria manifest lock is capped at 30 s on Windows only; the inquiry ledger alpha keeps refusing non-POSIX hosts. Two finally blocks that released an unacquired lock now release only what they acquired. SETUP docs state the best-effort Windows posture; no Windows CI job is added. Refs #845, #843, #844. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0131cZMWBPPeEFiqgEPFZ3X2 * fix(file_lock): interrupted attempts honour the deadline; pin adjudication wait policy (#845) Cross-model review round 1 (gpt-6-astra, xhigh): a persistent InterruptedError could retry past the bound; the Windows-shape test did not exercise adjudication's reader-waits / writer-does-not-wait policy; the adjudication contention message now names LockTimeout instead of BlockingIOError, recorded in the CHANGELOG rather than masked. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0131cZMWBPPeEFiqgEPFZ3X2 * refactor(file_lock): held() context manager, single BACKEND source, one fake msvcrt (#845) /simplify pass (four cleanup reviewers): the release-only-if-acquired invariant moves into file_lock.held() and review_criteria_binding / inquiry_branch_ledger use it; runtime branches key off BACKEND and SHARED_LOCKS_SUPPORTED is dropped; EINTR joins the retryable errno set and the unreachable EDEADLK entry goes; backend calls are deduplicated; all four consumers try the sibling import first so one module instance is shared; the Windows fake lives once in tests/fake_msvcrt.py; test scaffolding is folded into a lock_pair fixture and a parametrized wait test. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0131cZMWBPPeEFiqgEPFZ3X2 * fix(file_lock): keep lock acquisition and the guarded body in separate try blocks (#845) Cross-model review round 3 (gpt-6-astra, xhigh): wrapping the body in the same handler that translates LockTimeout meant a contended inner lock inside the body was reported as the outer manifest/passport lock failing. Both consumers now acquire in their own try block and release only after a successful acquire; held() is dropped from the helper. The subprocess test pins that a LockTimeout raised inside the binding body surfaces as itself. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0131cZMWBPPeEFiqgEPFZ3X2 * test(file_lock): let the body LockTimeout leave _locked() so the attribution check bites (#845) Cross-model review round 4: the inner LockTimeout was caught inside the binding body, so the erroneous outer translation would still have passed. Verified by mutation: restoring the outer translation fails this test. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0131cZMWBPPeEFiqgEPFZ3X2 * ci(673): whitelist scripts/test_file_lock.py as a non-consumer importer of the activity runtime (#845) The shared file-lock test imports adjudication_activity in a subprocess to exercise its lock backend under a fake msvcrt; it never reads or writes an activity store. The exact-owner whitelist is the lint's route for that. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0131cZMWBPPeEFiqgEPFZ3X2 --------- Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
6b7ee6dcae |
fix: Astra request compat, no-delegation citation transport, hedge/quota prompt repairs, audit provenance (#823–#826) (#827)
* fix: Astra request compatibility, no-delegation citation transport, hedge/quota prompt repairs, audit provenance (#823 #824 #825 #826) #823 — OpenAI request builders (smoke entrypoint + documented example) drop `temperature`, which GPT-6 Astra rejects; the per-model effort vocabulary lives in scripts/cross_model_verification/openai_effort_guard.sh, sourced by both, and an unsupported explicit Astra value fails before curl. Hermetic fake-curl test runs both surfaces. #824 — the contained Codex citation transport rejects effort=ultra with REASONING_EFFORT_REQUIRES_DELEGATION before detection/auth/tempdir/launch on both entry paths (codex-cli 0.153.4 defines ultra as the multiAgentMode replacement). Model-independent by design. #825 — hedging can no longer rescue an unsupported claim (writer recovery tree, CER fallback row, temporal rule 5 in writer + both compiler mirrors, writer contract D2); universal prose quotas in the writer, compilers, writing_quality_check.md, academic-paper/SKILL.md, and contract D6 become diagnostics subordinate to author/venue requirements. Audit inventory corrected in place; held-out seed evals/heldout/unsupported_claim_recovery (NOT_RUN) registered. #826 — run_codex_audit.sh pins gpt-6-astra/xhigh and records both in a new sidecar `model` block; claim_audit_pipeline binds an unknown judge identity to a run-local cache key (no cross-run reuse) instead of defaulting to gpt-5.5-xhigh. Review: /simplify (4 angles), codex gpt-5.6-sol xhigh 2 rounds (r1: 1 P1 + 1 P2 + 2 P3 fixed; r2: 0 P1/P2), /security-review 0 findings; all 102 spec-consistency steps + pytest manifest replayed locally. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BNKiXpdHx1T5F5RbXT2Ueu * docs(claude): record the #824 ultra reversal in the v3.21.2 key-additions line The v3.21.2 bullet still said the contained Codex citation transport accepts ultra; #824 on this branch rejects it as a delegation request. Add the reversal so the live instruction surface matches the transport. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01K7emV5r2aqZDJzAyYVuuDo --------- Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
0861bc8538 |
chore(models): align docs and guardrails to Claude Fable 5.1 and GPT-6 Astra (#819) (#820)
* chore(models): align docs and guardrails to Claude Fable 5.1 and GPT-6 Astra Read both vendor system cards in full and applied the model-update pass: - Claude Fable 5.1 named as the current frontier model (PERFORMANCE en/zh-TW with a dated list-price re-derivation; cross-model primary-row example). - gpt-6-astra listed as a provisional cross-model verifier on both transports and recommended under the #783 lifecycle policy; gpt-5.6-sol keeps its validated status on the ChatGPT-subscription citation transport. Entry-gate smoke PASS on that transport (2026-09-05, codex-cli 0.153.4). SETUP en/zh-TW example sets, id-status allowlist, bakeoff baseline text, and .claude/CLAUDE.md move together. - Codex citation transport: `ultra` joins the closed reasoning-effort set as a named constant, with a test pinning turn/start forwarding and fail-closed rejection of unknown values. - New guardrail: checkpoint decision provenance (authority in the pipeline state machine, operational mirror in the orchestrator), indexed as risk R11; both content-lock hashes updated in this commit. - Provider-side monitoring / safety interventions named as a never-a-verdict case in the cross-model doc and the degradation registry row. - Model tiering records that the resolved tier is the declared model; risk register R1/R4/R5/R6 residual gaps updated. - Harness-retirement audit for the model change: audits/harness-retirement-2026-09-model-update.md (0 prompt retirements, 4 applied currency fixes, 2 deferred, 8 keep-as-debt annotations). Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011sWwwG3oCbtL4cGhRsr5US * docs(changelog): align the model-update entries with the final text The [Unreleased] entries were written before the simplify pass moved the checkpoint-decision authority into the pipeline state machine, reused the existing transport-failure markers for provider-side interventions, and de-numbered the model-tiering note. Wording now matches the files. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011sWwwG3oCbtL4cGhRsr5US * test: scope the checkpoint-authority section out of the v3.6.7 orchestrator line budget The v3.6.7 Phase 6.6 budget test measures the orchestrator prompt minus every later independent extension, each with its own bounded cap. The new `## Checkpoint authority fidelity` section (13 lines) pushed the v3.6.7-attributed count to 652 against a 639 ceiling. Following the existing convention, the section gets its own measurement helper, an 18-line cap (5 lines of headroom), a dedicated test, and is subtracted from the historical budget. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011sWwwG3oCbtL4cGhRsr5US --------- Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
9469fc4d07 |
fix: fail check_surface_form_parity with an environment error when pyyaml is missing (#801 follow-up) (#803)
With the manifest present but pyyaml unimportable, _load_manifest returned None and main() misdiagnosed it as "manifest ... empty / null / non-mapping", pointing the reader at the wrong file. The import failure is now a distinct _YamlUnavailableError; the lint exits 1 naming pyyaml and the requirements-dev.txt remedy. Regression test pins the message (45 tests, all green; lint itself still passes). Also de-enumerate the stale "(PyYAML + jsonschema ...)" dependency parenthetical in docs/SETUP.md and docs/SETUP.zh-TW.md (both language files together, per bilingual-parity discipline). Claude-Session: https://claude.ai/code/session_013R81d1YwGvJAznkPKk9gNw Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
b6062c1401 |
feat: first Promotion Bakeoff run — gpt-5.6-sol validated for the codex subscription transport (#788)
* feat: first Promotion Bakeoff run — gpt-5.6-sol validated for the codex subscription transport (#787) Probe set: 30 refs (10 easy DOI-keyed journal articles; 10 hard: 3 arXiv, 2 DOI-less NeurIPS, 5 non-English; 10 fabrications), every real row resolver-confirmed same-day, every fabrication negative-checked. 180 same-day paired calls (30 x 3 repeats x 2 models), majority verdicts. Result: all five measures PASS with superiority — recall 1.00 vs 0.80, grounded completion 0.933 vs 0.900, p95 latency 26.5s vs 58.7s, zero guard misfires, false disagreement 0.00 = 0.00. Transport-qualified: gpt-5.6-sol stays provisional on the first-party API route (jq guards unexercised; allowlist unchanged). Report + probe-set sha256 under audits/; per-call index committed beside the probe set. Campaign side-product (transport): page-open webSearch items (action.type != "search") are skipped for binding instead of failing the stream (opened-page URLs still can never become bound sources), and DEVELOPER_INSTRUCTIONS requires an empty sources array for NOT_FOUND/NOT_SEARCHED. 52 transport tests green. Defective-tool run 1 archived unscored; three probe-row transcription errors were flagged MISMATCH by both models, independently re-verified, corrected, re-run. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Mo3QXHj2yzwNQ3VQKVEM2j * fix: narrow the page-open exemption to the observed action.type == "other" shape (#788 codex P2) An empty action object, unknown action type, or non-dict action on a webSearch item is stream-fatal again; only the observed page-open shape is skipped. Mutation test sweeps four bad shapes (52 -> 53 tests). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Mo3QXHj2yzwNQ3VQKVEM2j * fix: anchor the page-open exemption to the first-party closed WebSearchAction set Run-3 surfaced a third real shape ({"type": "openPage", "url": ...}) that the single-observation exemption rejected, tool-suppressing the baseline's measures (13 EVENT_STREAM_INVALID cells). The exempt set is now exactly the non-search members of the app-server protocol's closed WebSearchAction oneOf — {other, openPage, findInPage} plus the Responses-API spellings — verified against `codex app-server generate-json-schema` on 0.147.0. Unknown shapes stay stream-fatal (mutation sweep unchanged); 54 tests. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Mo3QXHj2yzwNQ3VQKVEM2j * docs: score preregistered run 4 as the gate result; runs 1-3 recorded as exploratory Run 4 (frozen fixture @ |
||
|
Edward Cheng-I Wu
|
075390a5c3 |
docs: cross-model recommendation surfaces follow generation currency (#784)
* docs: cross-model recommendation surfaces follow generation currency (#783) Recommendation decoupled from validation status: gpt-5.6-sol (current OpenAI flagship) becomes the lead OpenAI example while staying provisional — a dated lifecycle note records the flip carries no measurement claim; the Promotion Bakeoff remains the only route to validated. gpt-5.5 / gpt-5.5-pro demoted to validated previous-generation rows (measured bakeoff baseline unchanged). Gemini 3.1 Pro stays recommended (first-party check 2026-08-19: still Google's most capable Pro model). SETUP en/zh-TW quick-setup blocks updated in lockstep (parity lint green); the #630 guard's recommendation witness re-pinned to the new policy sentence with its mutation test updated in the same commit. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Mo3QXHj2yzwNQ3VQKVEM2j * fix: close codex round-1 P2s — evidence-ceiling wording + policy-body witness P2-1: "measured bakeoff baseline" overstated the evidence (no bakeoff run has ever been recorded); gpt-5.5 is the designated baseline, validated = allowlist status only. Reworded on all four surfaces (canonical doc, SETUP en/zh-TW, CHANGELOG). P2-2: the #630 recommendation-policy witness pinned only the heading; the guard now pins the two load-bearing body clauses (no-measurement-claim, bakeoff-only route to validated) with mutation tests for each (29 -> 31). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Mo3QXHj2yzwNQ3VQKVEM2j * fix: close codex round-2 P2 — superiority claim requires an observed measure The rewritten outcome bullet's "or operational benefit" branch let a measured-superiority claim rest on an unmeasured benefit; superiority now requires observed superiority on one of the five measures, and operational benefits are scoped to recommendation policy. Header no longer says "two distinct promotions" for what is now one promotion plus a claim rule. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Mo3QXHj2yzwNQ3VQKVEM2j * docs: close codex round-3 P2 — name the subscription-transport exception The citation-only codex subscription blocks (canonical + SETUP en/zh-TW) keep gpt-5.5 deliberately; a comment now names this as a transport-specific exception to the generation-currency recommendation and points at the codex smoke test before swapping ids. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Mo3QXHj2yzwNQ3VQKVEM2j --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
cdd48d916d |
docs: DATA_FLOWS.md — single map of network touchpoints + local stores (#758) (#770)
* docs: single data-flow map + DF-1..DF-3 coverage lint (#758) Add docs/DATA_FLOWS.md — one row per network touchpoint (trigger, payload class, recipient, credentials, off switch) and one row per local store (path, content, TTL, deletion), with an explicit scope statement (the Claude session itself is platform-governed; nothing publishes autonomously). Covers the four gate resolvers, the standalone Chinese-literature resolver (NOT in the gate), the consent-bound claim-standing discovery adapters, both cross-model transports (API and citation-only Codex subscription), the SessionStart update check, the manual smoke tests, and the v3.9.4 timeline bootstrap — the last one surfaced by the new lint itself on first run (it was absent from the #758 issue enumeration). Inbound links from README, SECURITY.md (in-scope exfiltration anchor), and THIRD_PARTY.md (core-suite vs third-party contrast). Lint (same-PR drift-point discipline): scripts/check_data_flows.py — DF-1 every non-test scripts/*.py importing a network module (AST scan, so no-call guards naming urllib.request in strings do not count) must be named on the map; DF-2 same for curl-invoking shell scripts; DF-3 README/SECURITY/THIRD_PARTY keep a rendered resolving inbound link (fences + HTML comments stripped with the semantics converged in the PR #768 review; consolidation into a shared helper is follow-up). 12 mutation tests; wired into spec-consistency.yml + pytest manifest. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * refactor: apply /simplify pass (4-agent, deduped) (#758) Doc: the four gate resolvers collapse into a 4-column sub-table under one shared trigger/payload/off-switch lead (the wide table kept only heterogeneous touchpoints); the exhaustiveness sentence is bounded to what DF-1/DF-2 actually detect (direct imports + curl; spawned-CLI and session-tooling paths held by review); "Nothing here publishes" now inherits POSITIONING.md and its not-a-runtime-guarantee qualifier; the subscription-free note is trimmed to its rationale; Related gains the SETUP bullet as the tunables authority. Coverage: docs/SETUP.md becomes the fourth DF-3-pinned inbound surface (pointer added in the cache section); the four translated READMEs mirror the README pointer; docs/DATA_FLOWS.md registers into check_spec_consistency.py relative-link validation. Lint: DF-1 module vocabulary rebuilt as the network subset of the no-call envelope FORBIDDEN_IMPORTS (deviations documented: dotted urllib.request/http.client instead of bare urllib/http; ssl excluded); scan is now recursive into scripts/ subpackages. Tunable constants in verification_cache.py gain update-both comments. Tests: the three hollow assert-baseline tests become real mutations (name-based test exemption, uncomment-curl, from-urllib idiom); recursive-scan and SETUP-surface tests added (17 total). Skipped with reason: endpoint-hostname lint (near-zero event rate, composed-URL false-fire risk); row-id shrink constant (review-owned per degradation-registry precedent); markdown-helper consolidation with check_control_availability.py (whichever PR merges second extracts the shared module — recorded in both PR bodies). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R1 findings — 8 P2 + 4 P3 (#758) Doc accuracy (6): Chinese-literature row rewritten (callable client, no CLI; PubMed path sends the required NCBI contact email + bibliographic search coordinates); codex-transport payload names citation_context (can contain unpublished manuscript text); update check documented as one curl transfer per 24 h with redirects and the ARS_UPDATE_CHECK_REMOTE_URL override; retraction-status SQLite cache added to local stores (caller-supplied path, 30-day stale threshold, no auto-expiry); discovery adapters credentials corrected (fixed User-Agent, resolver env keys not consumed); resolver payload narrowed to identifiers + title query strings; update-check state content corrected (state label + two version strings). Lint mis-pass/mis-fire (4): DF-1/DF-2 coverage now requires the full repo-relative path (basename-substring collision closed); DF-2 is recursive over scripts/ and hooks/, recognizes path-qualified curl, and masks quoted spans before the comment strip; DF-3 strips inline code spans before link extraction (a backticked link does not render). Five mutation tests added (22 total). The code-span rule is a divergence from check_control_availability.py to be carried over at the declared helper consolidation. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R2 findings — 4 P2 + 1 P3 (#758) - DF-2 scans command-substitution bodies BEFORE quote masking, so resp="$(curl ...)" — a real network call inside double quotes — fires (mutation test added; suite now genuinely 22, correcting the prior commit message which said 22 when 21 were collected). - Map gains the Codex audit wrapper row (scripts/run_codex_audit.sh: human/CI/hook-invoked only, sends deliverable + supporting file contents through the local Codex CLI login). - Update check re-bounded: at most one SUCCESSFUL check per 24 h; a failed attempt writes no state and may retry next session. - Cache TTL wording corrected: expiry is a cache miss, not deletion; expired rows persist until invalidated or the file is deleted. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: full comment lines execute nothing — DF-2 substitution scan (#758) The R2 command-substitution scan ran before any comment handling, so a full comment line containing $(curl ...) false-fired — surfaced by the codex R3 pass (timed out mid-review, but its transcript had already demonstrated the false fire). Comment-only lines are now skipped before the substitution scan; a $(curl) inside a trailing inline comment remains a documented accepted edge. Mutation test added (23 total). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R3 findings — command-position curl + image links + retention wording (#758) - DF-2 rebuilt around COMMAND POSITION: curl counts only as the first non-assignment token of a segment (pipes/separators/substitution openers), so `command -v curl` preflights and `echo curl` no longer false-fire; VAR=x curl still fires; wrapper-prefixed invocations (sudo/timeout) are documented accepted edges. - DF-3 link grammar excludes image syntax —  renders no anchor and cannot keep the acceptance criterion green. - Cache retention wording includes the overwrite path: expired rows persist until overwritten by re-verification, invalidated, or the file is deleted. - Three mutation tests added (26 total). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R4 finding — curl behind shell control words (#758) The command-position head-token scan now skips shell control words (if/elif/while/until/then/else/do/!/time/exec) before naming the head, so `if curl …; then` and `while ! curl …; do` fire while `if true; then` stays quiet. Two mutation tests (28 total). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R5 finding — option tokens after control words (#758) `time -p curl …` / option-bearing exec forms: the head scan now skips `-`-prefixed option tokens alongside assignments and control words, so the option cannot shadow the command head. Mutation test added (29 total). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R6 finding — harness spawned-CLI paths scoped out (#758) Three maintainer-only measurement scripts reach the network through locally authenticated CLIs (dispatch_e4_panel via claude -p, run_review_criteria_constructive_value via Codex, check_ranking_lift via gh api). They are not user-facing feature paths, so instead of diluting the touchpoint tables they are now an explicit named scope exclusion — the exhaustiveness claim no longer silently spans them. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R7 finding — boundary count wording (#758) "Two boundaries" became three after the R6 harness exclusion; the count is removed rather than maintained. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
43a02bf7e2 |
docs: per-channel control-availability matrix (#757) (#768)
* docs: per-channel control-availability matrix (#757) Add docs/CONTROL_AVAILABILITY.md — one row per enforcement mechanism, one column per install channel (plugin / skills copy / repo clone / Cowork / claude.ai Project / Claude Science / Pi), with honest active / conditional / absent cells, per-channel notes citing the existing scattered sources (README Requirements, SETUP methods, pi/README.md, hooks/run_guard.sh), and the guard's environment degradation table. Linked from README (Requirements + SETUP pointer) and SETUP (Installation methods intro). Evidence re-verified against the working tree: the channel set has grown past the six named in the issue (SETUP now also documents Cowork and the claude.ai 4a/4b split), so the matrix covers all seven documented channels. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * refactor: apply /simplify pass + add CA-1..CA-3 defrift lint (#757) Simplify round (4-agent review, findings deduped): - Drop the 'How to read an integrity claim' section (it had already drifted from the matrix) and the all-identical Upstream row; both replaced by one legend sentence and one paragraph. - Move channel-scoped caveats (Cowork / claude.ai / Claude Science / Pi) from per-cell footnotes into a 'Channel-wide limitation' column of the channel table; notes drop from 11 to 7. - De-drift row labels: no inline allowlist contents (canonical list is pinned by check_tools_allowlist.py), no exhaustive feature list, no hard-coded Claude Code minimum version (lives in SETUP Method 0). - README: single slimmed pointer (second link and both enumerations removed); pointer mirrored to the four translated READMEs and docs/SETUP.zh-TW.md. - Degradations table scoped to actual guard degradations (the slash-form version row was misfiled); registry backpointer added; guard-launcher registry registration split to #769. Lint (per the new-claim-surface-needs-lint-in-same-PR discipline): - scripts/check_control_availability.py — CA-1 links/anchors resolve, CA-2 every SETUP '### Method' heading reachable from the channel table, CA-3 README + SETUP inbound links pinned. Cell semantics stay owned by code review (degradation-registry posture). - 9 mutation tests; wired into spec-consistency.yml + pytest manifest (150 entries). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R1 findings — 4 P2 accuracy corrections (#757) - SessionStart announce/update-reminder row: Conditional, not Active (bash launcher on Windows needs Git Bash; reminder needs curl) — new note 8. - Cross-model note 6 no longer claims credentials+curl universally; the citation-only Codex subscription transport is named as the alternative transport behind the same consent boundary. - Pi channel limitation reworded: the wrapper supplies no orchestration but uses an installed Pi capability when available. - 'Enforcement mechanisms' claim language aligned to 'controls' in the purpose statement and all five README pointers (consistent with note 7's trust-based posture). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R2 findings — lint mis-pass cases + note-8 wording (#757) - CA-1 link grammar accepts optional quoted titles so a titled dead link cannot silently skip the check. - CA-2 counts only fragments on links whose resolved destination IS docs/SETUP.md — a same-slug anchor into a copied file no longer satisfies method coverage. - CA-3 checks resolved link destinations, not a filename substring — a label that keeps the filename while the target moves now fails. - Note 8: singular SessionStart hook (hooks.json defines one; the announce script runs the update check internally). - 3 new mutation tests pinning each mis-pass case (12 total). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R3 finding — commented-out markdown counts for nothing (#757) Strip HTML comments before extracting links and headings in all three invariants: a commented-out inbound link no longer satisfies CA-3, a commented-out SETUP method heading no longer demands CA-2 coverage, and a commented-out dead link no longer fires CA-1. Two mutation tests pin both directions (14 total). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R4 finding — GFM type-2 HTML-block semantics (#757) A line beginning with <!-- opens a raw-HTML block through the --> line (including trailing text on the closing line) or to EOF if unclosed; nothing on those lines renders. The comment stripper now models that line-level behavior before the inline-span strip, so a link after --> on a comment line cannot satisfy CA-3 and a dead link after an unclosed comment cannot fire CA-1. Two mutation tests pin both (16 total). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * test: fix R4 mutation scenario — line-start vs inline comment (#757) The previous commit's CA-3 HTML-block test inserted the comment mid-line (inside the blockquote), where GFM renders the link normally and the lint correctly stays quiet — the test scenario was wrong, not the lint. Replaced with a whole-line mutation that actually begins with <!--, and added the inline-comment symmetry case (link still renders → CA-3 satisfied). 17 tests green. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R5 finding — block-quoted HTML-block lines (#757) The type-2 HTML-block rule applies to block-quote content: the stripper now looks through leading '> ' markers before the line-start test, so '> <!-- note --> [link]' cannot satisfy CA-3. Deeper CommonMark laminations are declared out of scope in the docstring (the surfaces do not use them; a full parser is out of proportion for a maintainer-slip guard). 18 tests green. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R6 finding — repo-containment on CA-1 targets (#757) A relative link that resolves outside the repository root now fails CA-1 even when the host path exists — an over-deep ../.. slip must not be masked by an existing host file. Mutation test added (19 total). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R7 finding — fenced code excluded from extraction (#757) Fenced code regions render literally, and README/SETUP use fences today, so they are in-scope: a link inside a fence no longer satisfies CA-3, and a sample "### Method" heading inside a SETUP fence no longer demands CA-2 coverage. Fence stripping runs before the comment pass so a comment opener inside a fence stays literal. Two mutation tests (21 total). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc * fix: close codex R8 finding — CommonMark fence-length closing rule (#757) The fence stripper now tracks the opening run character and length: a closer must be a same-character run at least that long with only trailing whitespace, so a four-backtick fence demonstrating an inner triple-backtick block is no longer closed early. Mutation test added (22 total). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EosnA4RdUYgbF2KmZ1DTmc --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
54507ece69 |
fix: harden v3.20 review and integrity contracts (#747)
Closes #734 Closes #735 Closes #736 Closes #737 Closes #738 Closes #739 Closes #740 |
||
|
Edward Cheng-I Wu
|
f4d5c7a282 |
Add contained Codex subscription transport for citation integrity (#705)
Add a closed-schema, citation-only Codex subscription transport with structured app-server web-search provenance, hermetic containment, fail-visible degradation, producer wiring, tests, and CI guards. Supersedes #567 and closes #630. Co-authored-by: dcs-scd <dcs-scd@github.com> |
||
|
Edward Cheng-I Wu
|
b1b2f7af78 |
docs(setup): standing preferences via CLAUDE.md — documented design position (#634) (#649)
* docs(setup): standing preferences via CLAUDE.md — documented design position (#634) Adds the SETUP en/zh-TW section per the #634 decision record: template, the two honest limitations (model-judgment journal tiers; no output-directory setting by design), the search_strategy boundary, and the re-evaluation triggers. Env-flags table cross-references it. Docs only. Closes #634 Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_014JMdNxZFKxzm6koV5tB4gG * fix(docs): #649 codex round-1 — search_strategy owner is the Schema 2 Annotated Bibliography, not the RQ Brief; zh-TW clause-comma fixes Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_014JMdNxZFKxzm6koV5tB4gG --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
49e79a7c99 |
feat(commands): expose bare /ars-* aliases via explicit frontmatter name (#635)
* feat(commands): expose bare /ars-* aliases via explicit frontmatter name (#633) All 16 commands/ars-*.md declare name: == filename stem so Claude Code >= 2.1.216 plugin installs resolve the bare /ars-<mode> alias the SessionStart announce already advertises; the namespaced form stays canonical. Pre-2.1.216 tradeoff (name replaces the whole command name, bare-only surface, namespaced autocomplete lost) accepted and documented in SETUP en/zh-TW. New check_command_frontmatter_name.py lint (10 tests, manifest-registered) wired into the command-invariants workflow. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011LjhVWRdN55ugpgVzYQgNg * fix(lint): close YAML-equivalent name-key re-spelling channel (codex P2) A bare 'name:' line scan misses '"name":' / 'name :' variants, so a later YAML-equivalent duplicate could override the resolved command name while CI stays green. The lint now detects every name-like key variant and accepts exactly one canonical 'name: <stem>' line; all re-spellings fail closed. +4 mutation tests (quoted duplicate, spaced-colon duplicate, sole non-canonical spelling, quoted value). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011LjhVWRdN55ugpgVzYQgNg --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
276afa1d05 |
docs(setup): de-drift Method 4a description-length figure (440-842 → durable comparative form) (#564)
The hardcoded character range had drifted from reality (descriptions now run 566-986). Replace with the comparison that motivates the section and cannot go stale: each description exceeds claude.ai's 200-char upload cap while staying under Claude Code's 1,024-char allowance. Claude-Session: https://claude.ai/code/session_01EA3EvegVqKrkM62u7k9PHF Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
bbc0659272 |
docs(release): align all doc surfaces for v3.18.0 [skip-closes-check] (#560)
- CHANGELOG: promote [Unreleased] to [3.18.0] - 2026-07-18 (empty [Unreleased] kept for the coverage gate) - README + 4 translated READMEs (zh-TW/zh-CN/ja-JP/ko-KR): version badge to v3.18.0, new v3.18.0 changelog-summary section per language, Academic Pipeline heading to v3.18.0 - .claude/CLAUDE.md: new "v3.18 Key Additions" section (self-improvement survey integration: #547 scope advisory, #548 search-bounded novelty, #549 risk-stratified claim gate, #541 cache staleness + live re-validation, #540 cross-model reviewer track, #539 judge independence, #550 robustness evals, #542 docs anchor), skills table + Version Info to 3.18.0 / 2026-07-18 - academic-pipeline/SKILL.md: frontmatter + Version Info + last_updated to 3.18.0 / 2026-07-18; whole-file sha256 lock updated same-commit - .claude-plugin/plugin.json + marketplace.json: version to 3.18.0 - MODE_REGISTRY.md: Last updated to v3.18.0 (2026-07-18) - docs/ARCHITECTURE.md: title + pipeline component refs to v3.18.0 (historical v3.17.0 quality-gate row markers preserved) - docs/SETUP.md + SETUP.zh-TW.md: #541 env rows promoted Unreleased -> v3.18.0; stale #517 row corrected Unreleased -> v3.16.0 - scripts/check_spec_consistency.py + its unittest suite: version/date pins bumped to 3.18.0 / 2026-07-18 in the same commit (release-prep convention) Verified locally: check_version_consistency.py (plain + --tag v3.18.0), check_changelog_covers_merges.py, check_pipeline_boundary_semantics.py (+66 mutation tests), check_spec_consistency.py + unittest suite, check_setup_cross_model_parity.py, and the full pytest suite (3421 passed, 3 skipped, 1 xfailed) all green. Claude-Session: https://claude.ai/code/session_01Hi625UBf6GX7aeWJpJpSnE Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
f38d73f8c1 |
feat(cache): #541 staleness advisory + opt-in live re-validation for the citation-verification cache (#557)
* feat(cache): #541 staleness advisory + opt-in live re-validation for the citation-verification cache VerificationCache gains entry_age_days/stale_report (oldest-live-row age per citation; ARS_CACHE_STALE_ADVISORY_DAYS threshold, default 30, 0 disables, malformed falls back). Phase A A0.5 emits ADV-CACHE-<n> advisory rows for stale cache-served verifications (checkpoint display via the #547 template; never gates); ARS_CACHE_REVALIDATE=1 verifies HIGH-IMPACT-supporting stale references live (#549 tier composition). Invalidation cascade documented at the command + gate. Summary schema gains optional cache_age_days + cache_stale_advisory. 5 new pytest cases; SETUP en/zh-TW document both envs. External motivation: Ren et al. (2026, arXiv:2607.13104 §6.2.3) — scheduled review and attenuation; staleness/inconsistency as the signature failure mode of integrated external knowledge. Closes #541 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Hi625UBf6GX7aeWJpJpSnE * fix(cache): #541 codex round 1 — real cache-through wiring at the gate, robustness hardening, prose aligned to executable reality P1: verify_citation/verify_passport run cache-through by default (closes the #182 Delta-2 forward-decl): four resolvers route via new detailed contamination-signals wrappers (interoperable cache keys; queried_by kept on hits; bool-only signal layer preserved byte-for-byte via require_queried_by parametrization); CLI gains --no-cache; ARS_CACHE_REVALIDATE=1 re-verifies stale rows live per-row at the gate. P1: the HIGH-IMPACT tier coupling is dropped (tiers are assigned at E1, unavailable at A0; gate-level stale-row revalidation replaces it, cost documented). P2s: naive timestamps read as UTC, malformed rows are misses, clock-skew clamps to 0, advisory flag computed from the emitted rounded value; conservative oldest-live-row contract stated in the schema; checkpoint template vocabulary extended to ADV-CACHE-<n> (lock re-hashed); cascade made unconditional; §6.2.3 framed as design inference. CLI tests isolated from the real user cache (autouse tmp fixture). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Hi625UBf6GX7aeWJpJpSnE * fix(cache): #541 codex round 2 — omitted-argument sentinel defaults + typed hit validation for the gate P1: verify_citation/verify_passport use an _UNSET sentinel — omission constructs the default VerificationCache and derives revalidation from ARS_CACHE_REVALIDATE; explicit cache=None stays the live opt-out. The documented non-CLI gate path now caches/revalidates by default. P2: gate callers require typed hit payloads (matched: real bool, queried_by in {id,title}) so a malformed row can never launder into lookup_verified; the bool signal wrapper keeps its historical looser criteria. Gate/transport test suites gain autouse tmp-cache isolation (default-on can never touch the real user cache) + sentinel/env-revalidation/malformed-payload cases (145 green). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Hi625UBf6GX7aeWJpJpSnE --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
a4088f46c9 |
feat(plugin): #544 SessionStart update-available reminder for plugin installs
Closes #544 (follow-up to #543). New scripts/ars_update_check.sh + SessionStart announce integration: plugin installs behind main get a one-line session-start reminder pointing at /plugin update academic-research-skills. 24h cache, 3s network ceiling, ARS_UPDATE_CHECK=0 kill switch, every failure path silent, announce byte-identical when current / disabled / not a plugin install. Security: the version value that flows into SessionStart additionalContext is validated by an allow-known grammar (bounded numeric core + at most one recognized release marker, 32-char cap) at every entry point (remote, local, both cache fields) before it is cached or emitted, closing a prompt-injection / JSON-envelope-corruption path. Converged to 0 P1/P2 across three rounds of parallel codex (gpt-5.6-sol xhigh) + security-review + adversarial checks. 37-test hermetic suite (manifest id 544-update-reminder); SETUP en/zh-TW + CHANGELOG updated. Spec: docs/design/2026-07-18-544-update-reminder-spec.md. 🤖 Generated with [Claude Code](https://claude.com/claude-code) |
||
|
Edward Cheng-I Wu
|
48bde6a0b1 |
feat(tiering): #517 model tiering — judgment/execution split, two opt-in directions, default untouched (#520)
* feat(tiering): #517 model tiering — judgment/execution split, two opt-in directions, default untouched One tiering mechanism per the frozen #517 design (Lance Martin, "Cost effective harnesses with Fable", 2026-07-10): - Default (ARS_MODEL_TIERING unset): byte-equivalent — every agent stays model: inherit. Same opt-in philosophy as terminal_policies. - economy (frontier session): 13 execution-type agents dispatch one tier below the session model, floor Opus-class, never Sonnet; draft_writer flagged as the highest-savings / most quality-sensitive downgrade. - quality-boost (below-frontier session): judgment-type agents at the Stage 2.5/4.5 gates + final-review surfaces step up to the frontier tier; nothing is ever downgraded. - Both directions carry explicit no-op announcements; unknown values warn once and behave as unset. Tiers are relative positions, never hard-pinned model ids (v3.7.0 opus-floor lesson). Classification: 39 agents (26 judgment / 13 execution; issue header's 25/12 arithmetic corrected, membership unchanged) in scripts/model_tiering_manifest.json + the canonical table in shared/model_tiering.md, pinned to each other and to the *_agent.md files on disk by scripts/check_model_tiering.py (11 mutation tests; wired into spec-consistency.yml + local pytest manifest, 60 -> 61). Consumers: compact "## Model Tiering (#517, optional)" block in the four SKILL.md files; SETUP en/zh-TW env-table rows; prompt-caching guidance (same-stage calls reuse the same worker) in the canonical doc. No agent-file edits (sha256-locked bibliography_agent.md untouched), no schema change, no hook. Spec: docs/design/2026-07-12-517-model-tiering-spec.md All CI lints + 61-entry pytest manifest green locally. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FczvB1WJjfumdJTg9YyL3X * fix(tiering): #517 round-1 codex fixes — dispatch-as-subagent mechanism, exact-set lint, caching scope, wording unification Addresses the round-1 codex review (2 P1 + 6 P2; 2 no-change with reasons): - P1 routing reality: canonical doc now states the mechanism honestly — a tier is selectable only at subagent dispatch; when a direction applies to an inline role, the session dispatches it as a subagent pinned to the target tier (that IS the mechanism); impossible-dispatch falls open to inline-on-session-model with a one-line announcement. PERFORMANCE.md en/zh "no separate model routing layer" reconciled. - P2 caching-vs-default: same-worker guidance scoped to active directions; unset default byte-equivalent incl. dispatch shapes. - P2 semantics drift: quality-boost = jump TO the frontier (not one increment); warn-once wording unified; SETUP rows name the final-review surfaces. - P2 lint subset-only + hardcoded dirs: exact per-(tier,skill) token-set comparison (missing/extra/duplicate tokens, per-row counts, duplicate rows) + repo-wide stray sweep; tests 11 -> 15. - P2 stale verification record: 60 -> 61 manifest entries in the spec. - No-change (documented): the PR-body session link is the repo's established harness attribution convention (48 occurrences in recent main history); Lance Martin/Anthropic is published-source author attribution consistent with the repo's citation norm. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FczvB1WJjfumdJTg9YyL3X * fix(tiering): #517 round-2 codex fixes — dispatch-time tier resolver, claim-ref audit surface, Stage 3' roster, PERFORMANCE cost qualification - P1 tier resolver: new "Resolving a tier at dispatch time" section — the no-hard-pinning rule governs repo files; the dispatch call resolves the relative target from the runtime's own model information (alias or current-generation id, ephemeral only); unresolvable -> announced no-op. - P1 claim_ref_alignment_audit: quality-boost surface corrected to its real dispatch slot (opt-in Stage 4->5, ARS_CLAIM_AUDIT=1), removed from the 2.5/4.5 gate list; SKILL blocks + SETUP rows updated. - P2 Stage 3' roster: canonical text now matches ARCHITECTURE (narrow re-review team; judgment roles there = eic + editorial_synthesizer); caching example rewritten accordingly. - P2 PERFORMANCE en/zh: plugin-agent cost-unchanged claim qualified with the unset condition + economy behavior for plugin-exposed execution agents (report_compiler). - P2 PR body refreshed (dispatch-as-subagent wording, 15 tests). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FczvB1WJjfumdJTg9YyL3X * fix(tiering): #517 round-3 P2 residuals — caching example covers the full 3' narrow team; PR body audit surface - Caching example names the authoritative Stage 3' re-review roster (field_analyst + eic + editorial_synthesizer) and states the reuse rule is tier-independent — field_analyst is the economy-affected role. - PR body quality-boost bullet gains the opt-in Stage 4→5 claim-ref audit surface (was omitted after the round-2 canonical correction). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FczvB1WJjfumdJTg9YyL3X --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
6cee432e6a |
feat(cross-model): #518 gate hardening — risk-stratified sampling, blind disagreement checkpoints, id-status allowlist, promotion bakeoff (#519)
* feat(cross-model): #518 gate hardening — risk-stratified sampling, blind disagreement checkpoints, id-status allowlist, promotion bakeoff Four upgrades to shared/cross_model_verification.md + consumers, per the 2026-07-11 cross-model consult frozen in #518: 1. Integrity-gate cross-model sample: uniform random 30% (min 5 max 15) → risk stratification (100% of high-impact refs uncapped + 10% random remainder min 3 max 10; Stage 4.5 adds 100% of new/changed-claim refs + control sample). integrity_verification_agent updated in lockstep; results table gains a Tier column. 2. Blind disagreement checkpoints at the two irreversible decisions (research_architect design freeze; editorial_synthesizer final decision): independent structured cross-model decision, anchoring- blind, divergence = targeted rebuttal + user escalation — review trigger, never a vote, never averaged. Sprint-contract boundary kept closed (drivers never enter the scoring matrix). 3. "6th reviewer — Planned" retired, not deferred (counterproductive- conditions list matches documented anti-patterns); live mirrors (.claude/CLAUDE.md, raise_framework, SETUP en/zh-TW) drop the claim; historical release notes untouched. 4. Detection snippet separates endpoint routing from id trust: CROSS_MODEL_ID_STATUS=validated|provisional|unlisted with explicit unlisted-id warning; routing byte-identical (grounded precedence preserved). Plus §Promotion Bakeoff operationalizing the gpt-5.6-sol provisional→validated non-inferiority run (5 thresholds, smoke-test entry gate). Spec: docs/design/2026-07-12-518-cross-model-gate-hardening-spec.md All CI lints + 60-entry pytest manifest green locally. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FczvB1WJjfumdJTg9YyL3X * fix(cross-model): #518 round-2 codex fixes — NEW-CHANGED tier, primary pre-commitment, panel_size N, bakeoff reproducibility + validated-vs-default split Addresses all 6 findings from the round-1 codex review (gpt-5.6-sol, xhigh): - Stage 4.5 tier gap: 4-tier set (HIGH-IMPACT > NEW-CHANGED > CONTROL/RANDOM), mutually exclusive by precedence, one verification per reference; CONTROL replaces RANDOM at the final gate. - Cap contradiction: stale "max 15 survives" sentence removed; RANDOM/CONTROL capped at 10 each, 10% round-up rounding pinned. - Design-freeze primary decision: architect commits the same enum + drivers BEFORE the blind call, with criteria per enum value; blueprint Output Format gains a Design-Freeze Checkpoint Audit section. - Sprint-contract integration: editorial checkpoint is explicitly post-Step-3 (never extends the three-step arithmetic); input is the panel's panel_size N usable cards (5 full-mode / 2 methodology_focus), not a hardcoded five. - Bakeoff reproducibility: versioned, labeled, sha256-recorded probe-set fixture is a precondition; >=2/3 majority verdict with conservative 1-1-1-split handling defined. - Non-inferiority != default flip: full pass earns provisional->validated; the recommended default flips only with a stated superiority or operational-benefit reason. Spec + CHANGELOG updated to as-built; agents/ mirror re-copied. All CI lints + 60-entry pytest manifest green locally. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FczvB1WJjfumdJTg9YyL3X * fix(cross-model): #518 round-2 verify fixes — blindness leak, promotion wording, audit transport-error, cost bounds, SETUP tier coverage Addresses all 6 findings from the round-2 codex verify (1 P1 + 5 P2): - P1 blindness leak: the architect now commits its structured decision SEPARATELY from the blueprint and sends the cross-model a sanitized payload with the audit section stripped; the audit section is populated only after the comparison. - Two remaining canonical statements implying non-inferiority => default flip aligned to the validated-vs-default split. - Design-Freeze Checkpoint Audit gains unavailable / transport-error representations. - Cost table: bounded ranges replaced with a worked example (17 calls) + explicit no-fixed-upper-bound statement. - SETUP en/zh-TW feature rows now mention the Stage 4.5 NEW-CHANGED tier. - Spec Files-touched inventory adds the agents/ mirror; review-round annotations normalized (round 1 = first review, round 2 = verify). All CI lints + 60-entry pytest manifest green locally. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FczvB1WJjfumdJTg9YyL3X * fix(cross-model): #518 round-3 P2 residuals — audit N/A fields on transport failure, Limitations tier phrasing - Design-Freeze Checkpoint Audit: drivers/confidence gain explicit none/N/A values when the cross-model is unavailable; step 5 failure representation aligned to the schema. - Limitations §2 rewritten per-gate: RANDOM at Stage 2.5, NEW-CHANGED + CONTROL at Stage 4.5. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FczvB1WJjfumdJTg9YyL3X --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
93481c37ac |
feat(cross-model): list gpt-5.6-sol as provisional verifier + explicit reasoning-effort control (#515)
* feat(cross-model): list gpt-5.6-sol as provisional verifier + explicit reasoning-effort control Verified first-party against OpenAI's model page and GPT-5.6 guide (2026-07-11): Responses API + hosted web_search supported; effort enum none|low|medium|high|xhigh|max (default medium); premium is reasoning.mode "pro" on the standard slug (no -pro model id); standard rates identical to gpt-5.5. ARS-specific behavior is unvalidated, so gpt-5.5 stays the recommended default and gpt-5.6-sol is labeled provisional. New ARS_CROSS_MODEL_REASONING_EFFORT (default medium) makes the verifier's effort explicit; SETUP en/zh-TW mirror the example lines (parity lint green). New scripts/cross_model_smoke_test.sh is the manual live promotion gate. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FczvB1WJjfumdJTg9YyL3X * fix(cross-model): apply codex review P2s — preserve provider effort default, whole-word verdict match, fail-closed effort echo Unset ARS_CROSS_MODEL_REASONING_EFFORT now omits the reasoning field entirely (provider default preserved per model) instead of forcing medium; smoke test matches verdict tokens whole-word (UNVERIFIED can no longer false-pass as VERIFIED) and fails closed when a requested effort comes back without an echo. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FczvB1WJjfumdJTg9YyL3X * docs(changelog): fill in PR number #515 Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FczvB1WJjfumdJTg9YyL3X * fix(cross-model): smoke test enforces single-occurrence verdict token (codex re-review P2) A repeated verdict token now fails the gate instead of warning — the fixture prompt demands exactly one verdict, and a model that cannot comply with that instruction is itself a gate signal. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FczvB1WJjfumdJTg9YyL3X --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
cf973a8f7f |
fix(harness): retire expired scaffolds from 17-agent continuation audit (2 P1 + 9 P2) (#490)
* fix(harness): retire expired scaffolds from 17-agent continuation audit (#489) Apply the 11 user-accepted findings from audits/harness-retirement-2026-07-04.md (4-batch Fable 5 audit + codex cross-model challenge; 2 P1 + 9 P2): - B2-F01/B2-F02 (P1): deep-research socratic_mentor — delete stale 10/15-round constants contradicting the v3.0 40/60 + exploratory carve-out; merge the duplicated Auto-End machinery into one 6-condition authority - B3-F02 (P1, threshold=30 per user): academic-paper socratic_mentor — Auto-End Rules table becomes the single authority; conflicting 15-round bullets removed - B3-F01: 12-row question-template tables -> 1 canonical example per type - B3-F03: visualization Steps 2-6 narration -> pointer lines (6.5/6.6 untouched) - B1-F01/B1-F02: orchestrator — drop happy-path resume example; dispatch sub-list -> Checkpoint Confirmation Semantics pointer; add missing `view progress` row; fix duplicate step numbering - B1-F03: state_tracker — compress zero-information stage "2"/"3p" exemplars - B2-F03: report_compiler — trim basic citation few-shot, keep non-obvious rules - B4-F01: stale "prompt-level only / hook deferred to #134" enforcement sentence rewritten at 23 agents + 4 SKILL.md (guard shipped in #294); mirrors resynced - B4-F02: SETUP.md + zh-TW — gpt-5.4/Opus-4.8 lineup drift (6 lines each) User-rejected: B3-F04 (keyword mapping table stays), B4-F03 (field_analyst examples stay). 2026-06-10 F-007 closed as verified-no-rewrite-needed. Verification: run_ci_pytest_manifest 58/58 green; phase-boundary, write-scope, spec-consistency, collaboration-depth, mirror-sync lints green; personal-boundary scan 991 files / 0 violations. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01YBdGUAb184hmuiRiWAm4A9 * fix(harness): carry stop-here/change-settings aliases into semantics table (codex review P2) The B1-F02 dedup made the Checkpoint Confirmation Semantics table the single authority but dropped two aliases documented only in the deleted dispatch sub-list. Table is now a strict superset of the removed copy. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01YBdGUAb184hmuiRiWAm4A9 * refactor(harness): /simplify pass — tighten enforcement sentence, fix pointer label - S1/Efficiency: enforcement sentence body 39->35 words across 23 agents (+2 mirrors resynced); drops non-semantic slack (~16 tok/dispatch x 24 surfaces), keeps all five load-bearing facts and spec vocabulary (deterministic, rescope). Lint pins only the prefix marker — green. - S2: visualization Step 4 pointer label aligned to the actual heading "Figure Numbering and Captions (APA 7.0)". - ALT-1 (defrift lock: canonical-string pin in the phase-boundary lint + SETUP model-token parity) deliberately NOT done here — tracked as #491. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01YBdGUAb184hmuiRiWAm4A9 * chore(lint): bump bibliography_agent F2 baseline hash after B4-F01 sentence rewrite The v3.9.4 F2 ownership guard pins bibliography_agent.md by sha256 so any edit forces review. Reviewed: the B4-F01 apply changed only the Phase Boundary enforcement-status sentence; no M6 citation-provenance / M5 version-family / temporal logic moved, so the invariant holds. Baseline updated per the lint's documented procedure. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01YBdGUAb184hmuiRiWAm4A9 --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
8157a15b3b |
docs(release): align all doc surfaces for v3.14.0 (#481)
* docs(release): align all doc surfaces for v3.14.0 - CHANGELOG: new [3.14.0] entry (#480 Claude Science importability, #479 eval-comment renderer, #478/#477 prompt-debt retirement, #473 platform-port reminder, #464/#465/#469/#471 docs) + roll the 16-entry [Unreleased] backlog (code landed before the v3.13.0 tag) into the versioned record with a provenance note. - All five READMEs: version badge -> v3.14.0, pipeline version, translated v3.14.0 changelog entry, Claude Science import paragraph, five->six installation methods. (ja/ko/zh-CN entries machine-translated pending native review.) - docs/SETUP.md + SETUP.zh-TW.md: new Method 5 — Claude Science import (steps, snapshot semantics, what transfers / what does not). - .claude/CLAUDE.md: v3.14 Key Additions + suite version + table. - Version metadata: plugin.json, marketplace.json, academic-pipeline SKILL.md frontmatter/title/version table, CITATION.cff (version + date-released), POSITIONING.md citation, copilot-instructions. check_version_consistency passes (8 invariants); no eval/gate/skill behavior changes. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017jcH7gEkVTQ1baMbVedSZp * docs(release): bump spec-consistency pins + remaining v3.14.0 surfaces CI caught the surfaces the first pass missed: check_spec_consistency.py's expected-literal pins (+ its test fixtures), MODE_REGISTRY.md last-updated, academic-pipeline SKILL.md Version Info date, docs/ARCHITECTURE.md current-component markers (7 spots incl. mermaid node), and the ja README's full-width pipeline heading. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017jcH7gEkVTQ1baMbVedSZp * docs(i18n): apply agy review P3s — byte-identical zh-TW, restore "prompt" in ja/ko Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017jcH7gEkVTQ1baMbVedSZp --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
c086f19d87 |
docs(release): align all release docs to v3.11.0 (#182) [skip-closes-check] (#321)
* docs(release): align all release docs to v3.11.0 (#182) [skip-closes-check] The #182 citation-verification-gate feature code merged across PR-A/B/C1 but the release docs still said v3.10.0 and the CHANGELOG [Unreleased] body was empty. This aligns the 6 release-discipline invariants to v3.11.0. - CHANGELOG: convert [Unreleased] → [3.11.0] entry covering all five #182 deltas + C-V6 + lint (invariant 1). - README (en/zh-TW/zh-CN/ja-JP): version badge → v3.11.0; "Academic Pipeline" heading → v3.11.0; zh-TW adds a faithful v3.11 changelog entry; en/zh-CN/ja-JP sync version markers only (invariants 2/3/5). - .claude/CLAUDE.md: Skills Overview pipeline row → v3.11.0; Suite version → 3.11.0; Last Updated → 2026-06-04; add "v3.11 Key Additions" section. - docs/ARCHITECTURE.md: title + 6 pipeline-version markers → v3.11.0; evolution timeline gains v3.10.0 (a prior-release omission) + v3.11.0; Stage 5 FINALIZE gate now documents formatter rule 11 (v3.10 HIGH-BLOCK) + rule 12 (v3.11 citation_existence strict). - docs/SETUP{,.zh-TW}: add the citation-verification-cache section + the ARS_VERIFICATION_CACHE_PATH optional flag; correct the stale Opus 4.7 → 4.8 model recommendation. - docs/PERFORMANCE{,.zh-TW}: note that the #182 gate calls external APIs (no Claude token cost) + the cache amortization; correct Opus 4.7 → 4.8. - academic-pipeline/SKILL.md + plugin manifests: version → 3.11.0 (invariant 6). Only academic-pipeline tracks the suite version; the other three skill versions are unchanged. #182 was closed earlier (2026-06-02), so this does not re-close it. Version-consistency lint passes; full suite 2078 passed / 3 skipped. No tag in this PR — tagging is a separate, deliberate step. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * docs(release): sync version-mirror lint + fixtures + MODE_REGISTRY to v3.11.0 [skip-closes-check] The first PR-C2 commit aligned the human-facing docs but missed three mirrored surfaces that CI's spec-consistency lint guards — caught by the CI run, not the local check_version_consistency lint (which reads CHANGELOG dynamically; the spec lint hardcodes expected version strings). - MODE_REGISTRY.md: "Last updated: v3.10.0" → v3.11.0 (the 15th release doc, outside the original sweep). - scripts/check_spec_consistency.py: expected-string assertions bumped to v3.11.0 (badge / tag link / Suite version / "Academic Pipeline (vX)" heading / MODE_REGISTRY last-updated). Changelog-entry assertions ADD v3.11.0 while KEEPING the v3.10.0 history-guard line (these assert past entries still exist, so they must not be replaced). - scripts/test_check_spec_consistency.py: fixture templates parameterize the "Academic Pipeline (vX)" heading and carry both v3.11.0 + v3.10.0 changelog entries; aligned/stale fixtures and drift assertions rebased to v3.11.0 so the stale-badge regression tests still exercise real drift. - README.zh-CN.md + README.ja-JP.md: add the v3.11 changelog entry the lint now requires, each marked [machine-translated] pending native-contributor review (zh-TW was authored natively and is unmarked). Spec lint + its self-tests pass; version-consistency lint passes; full suite 2078 passed / 3 skipped. Mutation-checked: reverting a badge to v3.10.0 still fails the lint (not accept-all). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
57507ef7a0 |
docs(SETUP): fix Method 3 (Cowork) — zip upload, not ~/.claude/skills/ (#309)
Method 3 told users to symlink/copy the four skill folders into ~/.claude/skills/ and enable a toggle under Customize → Skills. That path is Claude Code's; Cowork does not read it, so the skills never appeared (reported in discussion #306). PR #307's "verified" root cause was incorrect. Verified first-hand against the on-disk skill store: Cowork loads skills uploaded via Settings → Capabilities → Skills, each as its own zip, registered by a server-issued skill ID. Uploading deep-research this way installs cleanly with the full description intact and /deep-research in the palette. - Rewrite Method 3 in SETUP.md + SETUP.zh-TW.md: zip-per-skill build + Settings upload + use + the orchestration trade-off vs Claude Code. - Fix a stale line claiming Method 3 keeps orchestration intact. - Fix QUICKSTART wording that listed Cowork next to ~/.claude/skills/. - EN/zh-TW H4 structure kept in parity. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
8316192358 |
docs(SETUP): add Cowork enable-skill step + fix command-palette wording (#307)
Method 3 (Cowork) registered the four skills via symlink but never said they must be enabled in Customize → Skills before they appear in the `/` command palette or get invoked. Verified by direct testing in Claude Desktop: a registered-but-not-enabled personal skill does not show in the palette; toggling it on (panel shows "Enabled") makes it appear. - Add step 6: enable each of the four skills in Customize → Skills. - Fix the "type / and select an available skill" line, which implied a freshly registered skill is selectable in the palette without enabling. - Mirror both changes in SETUP.zh-TW.md (H2 / step structure preserved). Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
ea87019b33 |
docs: signpost Codex users to ars-codex sibling distribution (#92)
Strengthen the existing ars-codex link by switching from a passive
"sibling distribution" mention to a direct "Using Codex CLI?" call-out
in README.md, mirror the same call-out into README.zh-TW.md (which was
missed in
|
||
|
Edward Cheng-I Wu
|
6006c5b5fa |
feat(v3.7.0 Phase 1 MVP): Claude Code plugin packaging — manifest + skills/ symlinks + dual-track install (3 codex rounds, 0 findings) (#68)
* docs(design): add v3.7.0 plugin packaging roadmap Phase 1 MVP (1-2h): .claude-plugin/plugin.json + marketplace.json + skills/ symlinks Phase 2 (1-2d): 10 slash commands + 7 agent council members + hooks (codex audit pipeline) Phase 3 (0.5d): marketplace publish + auto-update guidance + version bump checklist update Phase 4: traffic monitoring (clones / install ratio post-plugin) Verified facts via claude-code-guide agent: - Plugin auto-update: third-party plugins default OFF, user toggles per marketplace - Platform support: CC CLI / VS Code / JetBrains only — claude.ai web / cowork / API NO - Codex CLI: incompatible (separate AGENTS.md ecosystem); use academic-research-skills-codex fork - Min manifest: name + description (+ optional version) Isolation from concurrent v3.6.7 session: 0 path overlap, fetch-before-push. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(v3.7.0): patch roadmap with v3.6.7-derived updates + verified plugin spec facts Post-v3.6.7 ship retrospective patch (2026-05-05). Added "Update note" block at top of roadmap that supersedes parts of the original 2026-04-30 draft: - Verified plugin manifest spec facts (claude-code-guide query 2026-05-05): skills/ auto-discovery confirmed, symlinks resolve transparently per docs, min manifest = name only, marketplace.json shape uses source.{source: directory, path: ...} not plugins[].path. - Phase 2 hooks must wrap scripts/run_codex_audit.sh (already on main from v3.6.7 Phase 6.1), not reinvent codex audit. - Phase 2 agent council shrinks 7→3 (synthesis_agent / research_architect_agent / report_compiler_agent) per v3.6.7 Phase 6.7 inversion sweep. - Phase 2 plugin agent prompts must symlink or single-source from repo agents to avoid Pattern C3 attack surface re-introduction. - Phase 3 version sweep grows 7→9 categories (.claude-plugin/plugin.json added). - Bash 4+ hook safety reminder for macOS stock Bash 3.2 users. - Phase 0 inserted: spec verification + SKILL.md frontmatter check + CI baseline. Done as of this commit. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * feat(v3.7.0 Phase 1 MVP): Claude Code plugin packaging — manifest + skills/ symlinks + dual-track install docs Phase 1 of plugin packaging roadmap (docs/design/2026-04-30-ars-v3.7.0-plugin-packaging-roadmap.md). Lets users install ARS via `/plugin marketplace add Imbad0202/academic-research-skills` + `/plugin install academic-research-skills` while preserving the existing clone+symlink install path for the 4.3k clone+symlink user base. What lands: - .claude-plugin/plugin.json: name, version 3.7.0, description (covers v3.6.7 audit gate + v3.6.8 generator-evaluator contract), author, repo metadata, CC-BY-NC-4.0 license, 9 keywords. - .claude-plugin/marketplace.json: self-hosted marketplace shape per code.claude.com docs (source.{source: directory, path: .}); supersedes the speculative plugins[].path shape in the original roadmap. - skills/: relative symlinks pointing to the four sibling skill directories (deep-research, academic-paper, academic-paper-reviewer, academic-pipeline). Plugin loader resolves symlinks transparently per docs; legacy clone+symlink users unchanged. - docs/SETUP.md + docs/SETUP.zh-TW.md: new "Method 0: Claude Code Plugin (v3.7.0+, recommended)" section with auto-update guidance + platform-scope caveats (CC CLI/IDE only; claude.ai web/Claude for Work/API not supported; Codex CLI uses fork). - README.md + README.zh-TW.md: top of "Setup & installation" surfaces the one-line plugin install for Claude Code CLI/IDE users. Verified pre-commit: - python3 -c "import json; json.load(...)" parses both .claude-plugin/*.json. - ls -la skills/ shows 4 relative symlinks; readability check on each `<skill>/SKILL.md` resolves OK. - Repo regression: 742 pytest passed + 3 skipped (Bash 3.2 E2E gate) + 251 unittest = 993 green. - spec_consistency + version_consistency lint scripts pass. - No PII / hei-platform / HEEACT references in diff. Phase 2 (slash commands + agent mirrors + hooks wrapping run_codex_audit.sh) and Phase 3 (marketplace publish + auto-update) ship as separate PRs per roadmap. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(v3.7.0 Phase 1): codex review round 1 — 2 P1 + 1 P2 closed Round 1 trajectory: P1×2 + P2×1 + P3×0 → 0 (target). Closures (codex empirically validated via `claude plugin validate .`): - F-001 P1 marketplace_missing_owner: `claude plugin validate .` rejects with `owner: Invalid input: expected object, received undefined`. Earlier claude-code-guide spec query (2026-05-05) didn't surface owner as required. Fix: add top-level `"owner": {"name": "Cheng-I Wu", "url": "https://github.com/Imbad0202"}` per validator. - F-002 P1 marketplace_invalid_source_shape: `plugins[0].source` is a RELATIVE PATH STRING starting with `./`, not the object shape `{"source": "directory", "path": "."}` that earlier docs query reported. Validator output: `plugins.0.source: Invalid input`. Fix: source → "./". - F-003 P2 roadmap_documents_invalid_schema: roadmap update note's "Correct shape" block was based on the same erroneous spec. Fix: update note now shows the empirically-validated shape (owner + source as path string), explicitly marks the original Task 1.3 snippet below as obsolete. References: code.claude.com/docs/en/plugins-reference, code.claude.com/docs/en/plugin-marketplaces, plus empirical `claude plugin validate .` run by codex. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(v3.7.0 Phase 1): codex review round 2 — 1 P2 closed (validator clean) Round 2 trajectory: P2×1 → 0. Closure: - F-004 P2 plugin_update_command_misuse: docs used `/plugin marketplace update <name>` for manual plugin refresh, but that command only refreshes the marketplace source list — it doesn't update installed plugins. Correct command for refreshing an installed plugin is `/plugin update <name>`. Fix: docs/SETUP.md + docs/SETUP.zh-TW.md + roadmap Task 3.1 + Task 3.2 prose all updated to use `/plugin update` for plugin refresh, with a clarifying parenthetical about the marketplace-update distinction. Codex empirically validated this round: - `claude plugin validate .` passes - `claude plugin validate .claude-plugin/plugin.json` passes - `claude plugin tag --dry-run .` passes - 4 symlinks resolve to all SKILL.md files Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
Edward Cheng-I Wu
|
602af4bd36 |
fix(v3.6.5.2): mark claude.ai Method 4a not recommended for this Claude Code-native suite (#46)
* spec(v3.6.5.2): brief — claude.ai Method 4a scope clarification
Repositions v3.6.5.2 from the originally planned `SKILL.md` description
trim to a SETUP doc clarification. The trim direction was abandoned
because (1) only claude.ai's Custom Skills upload UI enforces 200 chars
(spec and Claude Code allow 1024), (2) ARS depends on Claude Code-only
Task / subagent dispatch and Material Passport cross-session handoffs
that are not part of the documented claude.ai Custom Skill runtime, and
(3) trimming would weaken Claude Code and Cowork routing — the platforms
ARS was built for — to unblock a path that delivers an unverified
partial fit.
Brief drafted, 5-round codex meta-review reaches zero P1 + zero P2 ship
gate. Patches the SETUP doc + CHANGELOG, no SKILL.md / agent / schema /
script / test changes.
* fix(v3.6.5.2): mark claude.ai Method 4a not recommended for this suite
SETUP.md Method 4 supersection rewritten:
- Method 4b (Project + GitHub integration) now presented first as the
recommended claude.ai path. Brings repo into Project knowledge for
reading and citation without losing fidelity.
- Method 4a (Custom Skill upload) marked not recommended for this suite
with rationale: ARS depends on Claude Code's Task / subagent dispatch
and Material Passport cross-session handoffs that are not part of the
documented claude.ai Custom Skill runtime, and trimming descriptions
to fit claude.ai's 200-char cap would weaken Claude Code and Cowork
routing in exchange for an unverified partial fit.
- Method 4a install commands kept in place for users who want to try
it anyway, framed as "if you want to try this path despite the
limitations". Zip-root packaging constraint reinstated.
- Prerequisites split per sub-method, citing Anthropic's plan
availability and code-execution requirement docs for Method 4a, and
Projects availability docs for Method 4b.
- Forward-looking caveat ("we have not run a live upload to characterise
this in detail") added so the recommendation is grounded in the
documented runtime scope, not in measured failure.
SETUP.zh-TW.md mirrors the English changes end-to-end.
CHANGELOG.md adds a v3.6.5.2 entry and reframes the v3.6.5.1 entry's
forward-promise of a description trim as "originally forecast … but
v3.6.5.2 instead documents Method 4a as not recommended."
No SKILL.md (frontmatter or body), no agent file, no schema, no script,
no test, no workflow, no version bump, no .claude/CLAUDE.md changes.
Refs: issue #44, brief commit
|
||
|
Edward Cheng-I Wu
|
aea9d24e2f |
fix(v3.6.5.1): SETUP doc correctness — install paths, claude.ai distinction, prereqs (#45)
* docs(SETUP): add known-broken warning banners to Method 3 + Method 4 Issue #44 (philpav, 2026-04-27): Cowork install does not register skills, claude.ai install does not work as documented. Codex audit on docs/SETUP.md found the root causes: - Method 3 instructions clone the whole repo to .claude/skills/ academic-research-skills/, burying every internal SKILL.md one level too deep — Cowork's discovery (.claude/skills/<skill>/ SKILL.md) never finds them. - Method 4 frames claude.ai's GitHub integration as a Skill install path, but it actually loads files into Project knowledge (static retrieval context), not the Skills system. Real Skill upload needs per-skill zip via Settings → Capabilities → Skills, and current description fields exceed the 200-char cap and would be rejected. Banner-first hotfix (Codex second-opinion advice 2026-04-27): add prominent WARNING banners on both methods immediately so users stop following broken instructions. Includes the 4-symlink Cowork workaround inline + cross-machine cp -R caveat. Full SETUP.md rewrite (v3.6.5.1) and SKILL.md description trim (v3.6.5.2) follow. This commit only adds banners; it does not yet rewrite the broken instructions below them. Users following the banner workaround will succeed; users skipping the banner and reading the original instructions will still fail (those will be fixed in v3.6.5.1). * spec(v3.6.5.1): SETUP fix implementation brief — 3-round codex meta-review Brief defines the constraint contract for the v3.6.5.1 SETUP.md rewrite. Given to a future Codex run as the non-negotiable spec sheet covering ARS-specific invariants, F1-F7 fix targets, authoritative Anthropic doc URLs, validation gate, and ship rules. Codex meta-review convergence (per feedback_codex_iterative_spec_review_to_zero.md discipline): - Round-1: 10 findings (3 P1 + 5 P2 + 2 P3); verdict: brief needs revisions before use. Issues spanned authoritative-source ordering (issue #44 reply was top, no inline content), validation realism (required Claude Code session unavailable in single-machine dev), §1.6 vs §4 sweep contradiction, banner removal commit authority, F2 wording self-contradiction, F3-F7 missing source URLs, F1 consolidation rule conflict with §8 drafter zone, §5 developer-environment risk, §6 ship gate too absolute. - Round-2: 2 cascade P2 (Fix-γ §1.6/§4 still conflicting; Fix-κ §6 ship gate too lenient with reject-with-rationale on P1+P2); verdict: do not ship. - Round-3: PASS (0 findings, 7/7 YES, ship verdict SHIP). Brief structure: - §1 Non-negotiable invariants (skill folder layout, discovery rule, Method 4 sub-method split, public repo Ruleset, authoritative source order with all 9 Anthropic doc URLs + verbatim issue #44 reply, allowed-edit set with sweep/edit boundary) - §2 F1-F7 fix specs (each with authoritative source URL) - §3 Banner removal (drafter prepares diff, human commits) - §4 Sweep checklist (sweep broad, edit only allowed) - §5 Validation gate (5.0 env protection mandatory; 5.1 required filesystem path-shape; 5.2 optional live discovery; 5.3 report) - §6 Codex review iteration + ship gate (P1/P2 zero non-negotiable, P3 either fixed or accepted-with-rationale-in-PR) - §7 PR body template - §8 Drafter judgement zone (explicit delegation list) Next: feed brief + audit raw + current SETUP.md to Codex as drafting input. Codex outputs SETUP.md rewrite. Iterate to ship gate. * fix(SETUP v3.6.5.1): rewrite Method 3 + Method 4 per implementation brief Issue #44 (philpav, 2026-04-27) — Cowork install does not register skills, claude.ai install does not work as documented. Rewrites docs/SETUP.md per the v3.6.5.1 implementation brief (commit |
||
|
Edward Cheng-I Wu
|
33f39f136b |
docs: refresh ARCHITECTURE / SETUP / PERFORMANCE to v3.6.5 (fast-follow) (#43)
* docs: refresh ARCHITECTURE.md to v3.6.5 (covers v3.5.1-v3.6.5)
ARCHITECTURE.md had drifted 5 minor releases behind main: titled v3.5.1
with skill versions v2.9.1 / v3.1.0 / v1.8.1 / v3.5.1. None of v3.5.1
reading-check probe, v3.6.2 sprint contract, v3.6.3 passport reset
boundary, v3.6.4 literature_corpus[] input port, or v3.6.5 corpus
consumers were documented. Codex /codex review on v3.6.5 PR-B did not
flag this because design doc §6.3 seven-touchpoint sweep didn't include
ARCHITECTURE.md. Adding to feedback_version_bump_sweep_checklist.md
backlog.
Sweep:
- Title v3.5.1 → v3.6.5
- All skill version pins synced: deep-research v2.9.1 → v2.9.2 (corpus
reader); academic-paper v3.1.0 → v3.1.1 (corpus reader); reviewer
v1.8.1 → v1.9.0; pipeline v3.5.1 → v3.6.5. Pins updated in §3 matrix
(every row), §6 mermaid skill graph, §9 modes table.
- §3 Stage 1 RESEARCH: bibliography_agent flagged as v3.6.5+ corpus
reader; socratic_mentor_agent flagged as v3.5.1 reading-check probe
layer (opt-in); Search Strategy artifact gains PRE-SCREENED block;
gate column gains corpus-first flow + 4 Iron Rules + F3/F4 mention.
- §3 Stage 2 WRITE: literature_strategist_agent flagged as v3.6.5+
corpus reader; Literature Search Report artifact gains PRE-SCREENED
block; gate column gains corpus-first flow.
- §3 Stage 3 REVIEW: editorial_synthesizer_agent flagged with v3.6.2
three-step mechanical protocol + forbidden-ops list; gate column
gains v3.6.2 Schema 13 sprint contract two-phase protocol with
paper-blind Phase 1 / paper-visible Phase 2 via <phase1_output>
data delimiter.
- New §5 "Material Passport literature_corpus[] Flow (v3.6.4 input
port + v3.6.5 consumers)": producer / passport / consumer mermaid
diagram, five-step shared flow narrative, four Iron Rules with full
text, PRE-SCREENED block description with F3/F4 reference, scope
exclusions (citation_compliance v3.6.6+ defer; source_pointer URI
resolution future). Section numbering shifted: old §5 → §6, §6 → §7,
§7 → §8, §8 → §9. §1 "How to read" updated accordingly + new bullet
for §5 corpus flow.
- §6 Skill Dependency Graph Shared block: added sprint_contract.schema,
reset_ledger_entry, literature_corpus_entry, rejection_log, and the
two reviewer contract templates (full / methodology_focus).
- §7 Quality Gates: 4 new entries — Reading-check probe (v3.5.1 opt-in),
Sprint Contract hard gate (v3.6.2), Passport reset boundary (v3.6.3
opt-in), Corpus consumer protocol (v3.6.5).
- §8 Timeline: title from "ARS v3.3.x incremental additions" to "ARS
evolution timeline"; v3.3.1-v3.3.6 collapsed into one entry (drift
fixes era); v3.5.1 / v3.6.2 / v3.6.3 / v3.6.4 / v3.6.5 each get
3-4 bullets.
Verification:
- check_spec_consistency.py exit 0 (markdown links re-validated)
- check_corpus_consumer_protocol.py exit 0
- pytest scripts/ 331 passed
Fast-follow on v3.6.5 (PR #42 merged main
|
||
|
Imbad0202
|
5c6de49840 |
docs: align model references with Opus 4.7 + adaptive thinking
- Upgrade all model IDs from claude-opus-4-6 to claude-opus-4-7 across performance guides, setup docs, cross-model verification protocol, reproducibility pattern, and example passport. - disclosure_mode_protocol: switch the example AI-usage statement from a hard-coded "Claude Opus 4.6" to a [MODEL_VERSION] placeholder, with a note instructing agents to pull the identifier from session metadata rather than bake in a version that drifts. - PERFORMANCE docs: drop the "extended-thinking configuration" phrasing in favor of a one-line note on Opus 4.7 adaptive thinking. - pipeline_orchestrator_agent: rewrite "Concise and clear, not verbose" (Opus 4.6 era negative framing) as a positive direction per the Opus 4.7 best-practices guide. All changes are documentation / example-config updates; test_check_repro_lock still passes (6/6) and no behavior change in the skills themselves. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
Imbad0202
|
a4420853c6 |
docs: streamline README, add SETUP/PERFORMANCE sub-docs (v3.3.6)
README cleanup rides on the ARCHITECTURE.md doc that PR #18 merged. Moves long-form setup, performance, and installation content into docs/ sub-pages so the READMEs read as entry points rather than manuals. New docs: - docs/SETUP.md + docs/SETUP.zh-TW.md — prerequisites, API key, Pandoc/ tectonic, cross-model verification (ARS_CROSS_MODEL), four install methods. - docs/PERFORMANCE.md + docs/PERFORMANCE.zh-TW.md — per-mode token budgets and recommended Claude Code settings (Agent Team / Ralph Loop / Skip Permissions). README changes: - Removed the ASCII pipeline diagram and the 16-point key-feature list (superseded by ARCHITECTURE.md §2/§3). - Skill Details section anchors version numbers and routes readers to ARCHITECTURE.md §3 for per-agent rosters; the four "### <Skill> (vX.Y)" headings are preserved so existing spec-consistency checks still pass. - Cross-Model Verification / Performance Notes / Prerequisites / Installation sections linked out instead of inlined. - Line counts: 816 → 459 EN (-44%), 791 → 444 ZH (-44%). Version bump to v3.3.6 across: - README.md + README.zh-TW.md (badge, release-tag link, changelog entry) - .claude/CLAUDE.md (suite version) - MODE_REGISTRY.md (last-updated line) - CHANGELOG.md (new [3.3.6] entry) CI adjustments: - scripts/check_spec_consistency.py: bumped expected version to v3.3.6 in both EN and ZH README checks; moved the DOCX contract assertions from the READMEs to the new docs/SETUP.* docs via a new check_setup_docs() step. No functional change to any skill. Pure documentation reorganization. |