docs: sync zh-TW to v3.0, add QUICKSTART, token estimates, LaTeX prereqs, fix broken links
- README.zh-TW.md: sync to v3.0 (features, cross-model section, token estimates,
LaTeX prereqs, v3.0 changelog, version badge, skill versions)
- .claude/CLAUDE.md: update to v3.0 (skills table, routing rules, version info)
- QUICKSTART.md: new 3-step quickstart guide with examples and mode selection table
- README.md: add token cost estimates per mode, add LaTeX/tectonic prerequisites,
fix broken Agent Team and Ralph Loop doc links
- README.zh-TW.md: fix same broken links
Addresses external review feedback: cognitive overload, missing quickstart,
token cost opacity, LaTeX dependency undocumented, zh-TW version drift, broken links.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-03 23:38:39 +08:00
# Quick Start
Get from zero to your first AI-assisted research in 3 steps.
## Step 1: Install
``` bash
# Install Claude Code
curl -fsSL https://claude.ai/install.sh | bash
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 ca21242), addressing 7 audit findings:
- F1 (P0) Method 3 install path: Option A (symlink) and Option B
(copy) now install each of the four skill folders separately
into ~/.claude/skills/, producing the discovery shape
~/.claude/skills/<skill-name>/SKILL.md instead of one level
deeper. Hard rule applied: every install command names all four
skills explicitly (no for loop) so readers cannot misuse.
Option A includes the cross-machine cp -R caveat for cloud-
synced ~/.claude/skills.
- F2 (P0) Method 4 conceptual split: Method 4a is the real
Custom Skill upload path (Settings → Capabilities → Skills,
one zip per skill, top-level entry <skill-name>/SKILL.md);
Method 4b is the GitHub-integration fallback knowledge mode
with explicit "this is NOT a Skill install" framing. Method 4a
is documented in advance with a known-blocker note that the
200-char description cap unblocks in v3.6.5.2.
- F3 (P1) Method 3 prerequisites: latest Desktop, internet,
app stays open during execution, folder permissions, paid
plan, org-level controls.
- F4 (P1) Method 4 prerequisites: split into 4a (no GitHub auth)
and 4b (Anthropic GitHub App authorisation, Team/Enterprise
owner enablement).
- F5 (P2) Cowork UI literal labels: mode selector (Chat / Cowork),
Tasks, Use an existing folder, Cowork Project. No paraphrasing.
- F6 (P2) Skill invocation: description-based routing as primary
mechanism + / command palette + + capability picker as explicit
UI surfaces. Removes the trigger-phrase-only oversimplification.
- F7 (P2) Method 4b directory table: scripts/ row added (adapter
+ validator workflows). Capacity rationale rewritten against
current Anthropic Project file limits (per-file 30 MB, count
unlimited).
Removes the d176578 banners (their purpose was harm reduction
during the patch interval; now superseded by correct instructions).
Codex iteration:
- Round-1: Codex drafted from brief + audit raw + current SETUP.md.
- Reviewer applied two manual fixes — Method 4a "immediately after"
loosened to "in a follow-up patch"; Method 4b table demoted .claude/
to Optional (avoids redundancy with recommended Project Instructions).
- Round-2: 2 P2 missing source URLs (Cowork Projects + Projects
static knowledge). Both added.
- Round-3: PASS (0 P1+P2). Ship verdict SHIP.
Sweep + validation + PR follow as separate commits.
* fix(QUICKSTART v3.6.5.1): align Step 1 install commands with new SETUP.md
QUICKSTART.md:14 had the same buried-skill bug as docs/SETUP.md
Method 3 — cloning into .claude/skills/academic-research-skills/
puts every SKILL.md one level too deep, so Claude Code/Cowork
never discover them.
Replaces with the four-skill symlink approach matching SETUP.md
Method 3 Option A (per F1 hard rule: explicit per-skill commands,
no for loop). Adds backpointer to SETUP.md for the copy-based
alternative and other install methods.
Sweep follow-up to commit ff935db (SETUP.md rewrite).
* fix(SETUP zh-TW v3.6.5.1): mirror English rewrite, full Method 1-4 update
Mirrors docs/SETUP.md (commit ff935db) into zh-TW. Rewrites Method
1-4 with the same F1-F7 fixes; preserves the existing Chinese
phrasing for non-install sections (literature_corpus adapters,
optional env flags, cross-model verification) where those are
not contradicted by the English audit.
Translation discipline (per user's writing rules):
- No em dashes (——) in Chinese prose (only inside shell/code comments).
- No formal-document connectors as paragraph openers (然而, 此外,
因此, 值得注意的是 etc.).
- No 吾人 archaic pronoun.
- No translation-style padding ("一個重要的考量是" etc.).
- Skill / Cowork / Project Knowledge use the established Chinese
terms (Skill, Cowork untranslated; Project 知識庫; 設定 → 功能 → Skills).
- UI literal labels kept in English with Chinese gloss in parens
(e.g. "Use an existing folder"（使用既有資料夾）) so users can
match what they click in the actual UI.
- agentic kept untranslated, consistent with ARS English terminology.
- All 9 Anthropic doc URLs from English version preserved.
- Bash code blocks identical to English (4 skills explicit, no for loop).
Codex iteration:
- Round-1: codex drafted Chinese mirror under the 8 writing-rule constraints.
- Reviewer applied 5 polish fixes:
* Removed "請請" typo + retired 控制項 → 設定 (Microsoft .NET wording)
* Owner-level connector enablement properly localised
* "not artificially capped at 200" rephrased to clearer Chinese logic
- Round-2: PASS (0 P1+P2). 1 P3 accepted (em dashes only in shell
comments, not Chinese prose).
Sweep follow-up to commits ff935db (SETUP.md) + 3ece187 (QUICKSTART).
* docs(v3.6.5.1): CHANGELOG entry for SETUP doc correctness patch
Doc-only patch summarising the F1-F7 fixes already shipped in commits
ff935db (SETUP.md), 3ece187 (QUICKSTART.md), and 06f6b71 (SETUP.zh-TW.md):
- Method 3 install paths fixed (per-skill folders into ~/.claude/skills/)
- Method 4 split into 4a (Custom Skill upload) and 4b (Project + GitHub)
- Method 3 + Method 4 prerequisites expanded
- Cowork UI terminology refreshed
- Skill invocation framing clarified (description-based + / + + UI)
- Method 4 directory table includes scripts/; capacity guidance refreshed
Issue #44 (philpav) reports the SETUP problems on Cowork and claude.ai;
this patch fixes the doc, and v3.6.5.2 will trim the four SKILL.md
descriptions to clear the claude.ai upload blocker.
2026-04-27 21:18:27 +08:00
# Clone this repo somewhere stable
git clone https://github.com/Imbad0202/academic-research-skills.git ~/academic-research-skills
# Install each of the four skills into your project's .claude/skills/
docs: sync zh-TW to v3.0, add QUICKSTART, token estimates, LaTeX prereqs, fix broken links
- README.zh-TW.md: sync to v3.0 (features, cross-model section, token estimates,
LaTeX prereqs, v3.0 changelog, version badge, skill versions)
- .claude/CLAUDE.md: update to v3.0 (skills table, routing rules, version info)
- QUICKSTART.md: new 3-step quickstart guide with examples and mode selection table
- README.md: add token cost estimates per mode, add LaTeX/tectonic prerequisites,
fix broken Agent Team and Ralph Loop doc links
- README.zh-TW.md: fix same broken links
Addresses external review feedback: cognitive overload, missing quickstart,
token cost opacity, LaTeX dependency undocumented, zh-TW version drift, broken links.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-03 23:38:39 +08:00
cd /path/to/your/project
mkdir -p .claude/skills
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 ca21242), addressing 7 audit findings:
- F1 (P0) Method 3 install path: Option A (symlink) and Option B
(copy) now install each of the four skill folders separately
into ~/.claude/skills/, producing the discovery shape
~/.claude/skills/<skill-name>/SKILL.md instead of one level
deeper. Hard rule applied: every install command names all four
skills explicitly (no for loop) so readers cannot misuse.
Option A includes the cross-machine cp -R caveat for cloud-
synced ~/.claude/skills.
- F2 (P0) Method 4 conceptual split: Method 4a is the real
Custom Skill upload path (Settings → Capabilities → Skills,
one zip per skill, top-level entry <skill-name>/SKILL.md);
Method 4b is the GitHub-integration fallback knowledge mode
with explicit "this is NOT a Skill install" framing. Method 4a
is documented in advance with a known-blocker note that the
200-char description cap unblocks in v3.6.5.2.
- F3 (P1) Method 3 prerequisites: latest Desktop, internet,
app stays open during execution, folder permissions, paid
plan, org-level controls.
- F4 (P1) Method 4 prerequisites: split into 4a (no GitHub auth)
and 4b (Anthropic GitHub App authorisation, Team/Enterprise
owner enablement).
- F5 (P2) Cowork UI literal labels: mode selector (Chat / Cowork),
Tasks, Use an existing folder, Cowork Project. No paraphrasing.
- F6 (P2) Skill invocation: description-based routing as primary
mechanism + / command palette + + capability picker as explicit
UI surfaces. Removes the trigger-phrase-only oversimplification.
- F7 (P2) Method 4b directory table: scripts/ row added (adapter
+ validator workflows). Capacity rationale rewritten against
current Anthropic Project file limits (per-file 30 MB, count
unlimited).
Removes the d176578 banners (their purpose was harm reduction
during the patch interval; now superseded by correct instructions).
Codex iteration:
- Round-1: Codex drafted from brief + audit raw + current SETUP.md.
- Reviewer applied two manual fixes — Method 4a "immediately after"
loosened to "in a follow-up patch"; Method 4b table demoted .claude/
to Optional (avoids redundancy with recommended Project Instructions).
- Round-2: 2 P2 missing source URLs (Cowork Projects + Projects
static knowledge). Both added.
- Round-3: PASS (0 P1+P2). Ship verdict SHIP.
Sweep + validation + PR follow as separate commits.
* fix(QUICKSTART v3.6.5.1): align Step 1 install commands with new SETUP.md
QUICKSTART.md:14 had the same buried-skill bug as docs/SETUP.md
Method 3 — cloning into .claude/skills/academic-research-skills/
puts every SKILL.md one level too deep, so Claude Code/Cowork
never discover them.
Replaces with the four-skill symlink approach matching SETUP.md
Method 3 Option A (per F1 hard rule: explicit per-skill commands,
no for loop). Adds backpointer to SETUP.md for the copy-based
alternative and other install methods.
Sweep follow-up to commit ff935db (SETUP.md rewrite).
* fix(SETUP zh-TW v3.6.5.1): mirror English rewrite, full Method 1-4 update
Mirrors docs/SETUP.md (commit ff935db) into zh-TW. Rewrites Method
1-4 with the same F1-F7 fixes; preserves the existing Chinese
phrasing for non-install sections (literature_corpus adapters,
optional env flags, cross-model verification) where those are
not contradicted by the English audit.
Translation discipline (per user's writing rules):
- No em dashes (——) in Chinese prose (only inside shell/code comments).
- No formal-document connectors as paragraph openers (然而, 此外,
因此, 值得注意的是 etc.).
- No 吾人 archaic pronoun.
- No translation-style padding ("一個重要的考量是" etc.).
- Skill / Cowork / Project Knowledge use the established Chinese
terms (Skill, Cowork untranslated; Project 知識庫; 設定 → 功能 → Skills).
- UI literal labels kept in English with Chinese gloss in parens
(e.g. "Use an existing folder"（使用既有資料夾）) so users can
match what they click in the actual UI.
- agentic kept untranslated, consistent with ARS English terminology.
- All 9 Anthropic doc URLs from English version preserved.
- Bash code blocks identical to English (4 skills explicit, no for loop).
Codex iteration:
- Round-1: codex drafted Chinese mirror under the 8 writing-rule constraints.
- Reviewer applied 5 polish fixes:
* Removed "請請" typo + retired 控制項 → 設定 (Microsoft .NET wording)
* Owner-level connector enablement properly localised
* "not artificially capped at 200" rephrased to clearer Chinese logic
- Round-2: PASS (0 P1+P2). 1 P3 accepted (em dashes only in shell
comments, not Chinese prose).
Sweep follow-up to commits ff935db (SETUP.md) + 3ece187 (QUICKSTART).
* docs(v3.6.5.1): CHANGELOG entry for SETUP doc correctness patch
Doc-only patch summarising the F1-F7 fixes already shipped in commits
ff935db (SETUP.md), 3ece187 (QUICKSTART.md), and 06f6b71 (SETUP.zh-TW.md):
- Method 3 install paths fixed (per-skill folders into ~/.claude/skills/)
- Method 4 split into 4a (Custom Skill upload) and 4b (Project + GitHub)
- Method 3 + Method 4 prerequisites expanded
- Cowork UI terminology refreshed
- Skill invocation framing clarified (description-based + / + + UI)
- Method 4 directory table includes scripts/; capacity guidance refreshed
Issue #44 (philpav) reports the SETUP problems on Cowork and claude.ai;
this patch fixes the doc, and v3.6.5.2 will trim the four SKILL.md
descriptions to clear the claude.ai upload blocker.
2026-04-27 21:18:27 +08:00
ln -s ~/academic-research-skills/deep-research .claude/skills/deep-research
ln -s ~/academic-research-skills/academic-paper .claude/skills/academic-paper
ln -s ~/academic-research-skills/academic-paper-reviewer .claude/skills/academic-paper-reviewer
ln -s ~/academic-research-skills/academic-pipeline .claude/skills/academic-pipeline
docs: sync zh-TW to v3.0, add QUICKSTART, token estimates, LaTeX prereqs, fix broken links
- README.zh-TW.md: sync to v3.0 (features, cross-model section, token estimates,
LaTeX prereqs, v3.0 changelog, version badge, skill versions)
- .claude/CLAUDE.md: update to v3.0 (skills table, routing rules, version info)
- QUICKSTART.md: new 3-step quickstart guide with examples and mode selection table
- README.md: add token cost estimates per mode, add LaTeX/tectonic prerequisites,
fix broken Agent Team and Ralph Loop doc links
- README.zh-TW.md: fix same broken links
Addresses external review feedback: cognitive overload, missing quickstart,
token cost opacity, LaTeX dependency undocumented, zh-TW version drift, broken links.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-03 23:38:39 +08:00
```
2026-06-03 19:08:23 +08:00
Each skill must sit at `.claude/skills/<skill-name>/SKILL.md` for Claude Code to discover it. See [docs/SETUP.md ](docs/SETUP.md ) for the copy-based alternative, global `~/.claude/skills/` install, and the other installation methods (Claude Code plugin, Cowork via zip upload, claude.ai). Note that Cowork and claude.ai do not read `~/.claude/skills/` — they install skills through their own settings upload, not this path.
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 ca21242), addressing 7 audit findings:
- F1 (P0) Method 3 install path: Option A (symlink) and Option B
(copy) now install each of the four skill folders separately
into ~/.claude/skills/, producing the discovery shape
~/.claude/skills/<skill-name>/SKILL.md instead of one level
deeper. Hard rule applied: every install command names all four
skills explicitly (no for loop) so readers cannot misuse.
Option A includes the cross-machine cp -R caveat for cloud-
synced ~/.claude/skills.
- F2 (P0) Method 4 conceptual split: Method 4a is the real
Custom Skill upload path (Settings → Capabilities → Skills,
one zip per skill, top-level entry <skill-name>/SKILL.md);
Method 4b is the GitHub-integration fallback knowledge mode
with explicit "this is NOT a Skill install" framing. Method 4a
is documented in advance with a known-blocker note that the
200-char description cap unblocks in v3.6.5.2.
- F3 (P1) Method 3 prerequisites: latest Desktop, internet,
app stays open during execution, folder permissions, paid
plan, org-level controls.
- F4 (P1) Method 4 prerequisites: split into 4a (no GitHub auth)
and 4b (Anthropic GitHub App authorisation, Team/Enterprise
owner enablement).
- F5 (P2) Cowork UI literal labels: mode selector (Chat / Cowork),
Tasks, Use an existing folder, Cowork Project. No paraphrasing.
- F6 (P2) Skill invocation: description-based routing as primary
mechanism + / command palette + + capability picker as explicit
UI surfaces. Removes the trigger-phrase-only oversimplification.
- F7 (P2) Method 4b directory table: scripts/ row added (adapter
+ validator workflows). Capacity rationale rewritten against
current Anthropic Project file limits (per-file 30 MB, count
unlimited).
Removes the d176578 banners (their purpose was harm reduction
during the patch interval; now superseded by correct instructions).
Codex iteration:
- Round-1: Codex drafted from brief + audit raw + current SETUP.md.
- Reviewer applied two manual fixes — Method 4a "immediately after"
loosened to "in a follow-up patch"; Method 4b table demoted .claude/
to Optional (avoids redundancy with recommended Project Instructions).
- Round-2: 2 P2 missing source URLs (Cowork Projects + Projects
static knowledge). Both added.
- Round-3: PASS (0 P1+P2). Ship verdict SHIP.
Sweep + validation + PR follow as separate commits.
* fix(QUICKSTART v3.6.5.1): align Step 1 install commands with new SETUP.md
QUICKSTART.md:14 had the same buried-skill bug as docs/SETUP.md
Method 3 — cloning into .claude/skills/academic-research-skills/
puts every SKILL.md one level too deep, so Claude Code/Cowork
never discover them.
Replaces with the four-skill symlink approach matching SETUP.md
Method 3 Option A (per F1 hard rule: explicit per-skill commands,
no for loop). Adds backpointer to SETUP.md for the copy-based
alternative and other install methods.
Sweep follow-up to commit ff935db (SETUP.md rewrite).
* fix(SETUP zh-TW v3.6.5.1): mirror English rewrite, full Method 1-4 update
Mirrors docs/SETUP.md (commit ff935db) into zh-TW. Rewrites Method
1-4 with the same F1-F7 fixes; preserves the existing Chinese
phrasing for non-install sections (literature_corpus adapters,
optional env flags, cross-model verification) where those are
not contradicted by the English audit.
Translation discipline (per user's writing rules):
- No em dashes (——) in Chinese prose (only inside shell/code comments).
- No formal-document connectors as paragraph openers (然而, 此外,
因此, 值得注意的是 etc.).
- No 吾人 archaic pronoun.
- No translation-style padding ("一個重要的考量是" etc.).
- Skill / Cowork / Project Knowledge use the established Chinese
terms (Skill, Cowork untranslated; Project 知識庫; 設定 → 功能 → Skills).
- UI literal labels kept in English with Chinese gloss in parens
(e.g. "Use an existing folder"（使用既有資料夾）) so users can
match what they click in the actual UI.
- agentic kept untranslated, consistent with ARS English terminology.
- All 9 Anthropic doc URLs from English version preserved.
- Bash code blocks identical to English (4 skills explicit, no for loop).
Codex iteration:
- Round-1: codex drafted Chinese mirror under the 8 writing-rule constraints.
- Reviewer applied 5 polish fixes:
* Removed "請請" typo + retired 控制項 → 設定 (Microsoft .NET wording)
* Owner-level connector enablement properly localised
* "not artificially capped at 200" rephrased to clearer Chinese logic
- Round-2: PASS (0 P1+P2). 1 P3 accepted (em dashes only in shell
comments, not Chinese prose).
Sweep follow-up to commits ff935db (SETUP.md) + 3ece187 (QUICKSTART).
* docs(v3.6.5.1): CHANGELOG entry for SETUP doc correctness patch
Doc-only patch summarising the F1-F7 fixes already shipped in commits
ff935db (SETUP.md), 3ece187 (QUICKSTART.md), and 06f6b71 (SETUP.zh-TW.md):
- Method 3 install paths fixed (per-skill folders into ~/.claude/skills/)
- Method 4 split into 4a (Custom Skill upload) and 4b (Project + GitHub)
- Method 3 + Method 4 prerequisites expanded
- Cowork UI terminology refreshed
- Skill invocation framing clarified (description-based + / + + UI)
- Method 4 directory table includes scripts/; capacity guidance refreshed
Issue #44 (philpav) reports the SETUP problems on Cowork and claude.ai;
this patch fixes the doc, and v3.6.5.2 will trim the four SKILL.md
descriptions to clear the claude.ai upload blocker.
2026-04-27 21:18:27 +08:00
docs: sync zh-TW to v3.0, add QUICKSTART, token estimates, LaTeX prereqs, fix broken links
- README.zh-TW.md: sync to v3.0 (features, cross-model section, token estimates,
LaTeX prereqs, v3.0 changelog, version badge, skill versions)
- .claude/CLAUDE.md: update to v3.0 (skills table, routing rules, version info)
- QUICKSTART.md: new 3-step quickstart guide with examples and mode selection table
- README.md: add token cost estimates per mode, add LaTeX/tectonic prerequisites,
fix broken Agent Team and Ralph Loop doc links
- README.zh-TW.md: fix same broken links
Addresses external review feedback: cognitive overload, missing quickstart,
token cost opacity, LaTeX dependency undocumented, zh-TW version drift, broken links.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-03 23:38:39 +08:00
## Step 2: Launch
``` bash
claude
```
## Step 3: Start researching
Tell Claude what you want to do. It will automatically pick the right skill and mode.
### Example: Guided research (Socratic mode)
```
You: "I have a vague idea about AI's impact on higher education quality assurance,
but I'm not sure how to frame the research question. Can you guide me?"
```
Claude will enter Socratic mode — asking questions to help you clarify your thinking, not giving you answers directly. After 5-15 rounds of dialogue, you'll have a focused research question and methodology direction.
### Example: Write a paper
```
You: "Help me write a paper about the impact of declining birth rates
on private universities in Taiwan"
```
### Example: Review an existing paper
```
You: "Review this paper" (then paste or attach the paper)
```
### Example: Full pipeline (research → write → review → revise → publish)
```
You: "I want to produce a complete research paper about how agentic AI
is reshaping student learning outcome measurement"
```
This triggers the full 10-stage pipeline. Budget ~$4-6 in API costs and 2-4 hours of collaborative work.
## Which mode should I use?
| I want to... | Use this |
|-------------|----------|
| Explore a vague idea | `deep-research` socratic mode — just describe your interest |
| Get a quick literature summary | `deep-research` quick mode |
| Do a systematic review (PRISMA) | `deep-research` systematic-review mode |
| Write a paper from scratch | `academic-paper` full mode |
| Plan a paper chapter by chapter | `academic-paper` plan mode |
| Get my paper reviewed | `academic-paper-reviewer` full mode |
| Do everything end-to-end | `academic-pipeline` — say "I want a complete research paper" |
## What's next?
- [Full README ](README.md ) — all features, modes, installation options, and changelog
- [中文版 ](README.zh-TW.md ) — Traditional Chinese version
- [Pipeline showcase ](examples/showcase/ ) — real artifacts from a complete pipeline run
