skills/boshu2__agentops
1
0
Fork 0
mirror of https://github.com/boshu2/agentops.git synced 2026-09-14 15:08:13 +08:00
Code Issues Packages Projects Releases Wiki Activity
Files
e1fae0dae63d73be508bb8348a12106df043cab0
boshu2__agentops/docs/contracts/ubiquitous-language.md
T

59 lines
3.9 KiB
Markdown
Raw Normal View History

refactor: cut AgentOps to a single-pass evidence loop
2026-07-14 22:01:50 -04:00
# Ubiquitous language
Align AgentOps around its operations-layer identity (#1051) Executes docs/plans/2026-08-07-agentops-operations-layer-alignment.md: AgentOps is the operations layer for agentic engineering; the federated integration graph is the topology, the semantic work-and-proof protocol is the contract, and RPI is the standard one-experiment traversal. Retires the ao flywheel command family and all knowledge-flywheel product state, tombstones the seven-move operating-loop workflow, narrows ao init and the .agents state writers to declared destinations, renames the core architecture page to rpi-traversal.md with a compatibility redirect, aligns AGENTS.md, 25 skills, public and package copy, regenerates every owned projection, and strengthens the conformance gates with planted-negative proofs. Both the alignment subject and the follow-up gate-bookkeeping commit carry fresh author-distinct validation PASS verdicts with empty not_checked scope. Test-Removal-Reason: dead knowledge-flywheel and session-store surfaces were deleted with their tests (operations-layer alignment)
2026-08-07 18:37:03 -04:00
## Product architecture
| Term | Meaning |
|---|---|
| Operations layer | The product category: the portable layer that makes heterogeneous agentic engineering systems interoperate semantically. |
| Federated integration graph | The topology: caller-owned intent, source systems, agents, factories, checks, and judgments remain separate nodes joined by typed handoffs. |
| Semantic work-and-proof protocol | The interoperability contract: exact intent, exact subject, evidence, fresh judgment, and honest outcomes. |
Make the engineering harness lean and add optional topic memory (#1116) RPI now owns the authorized outcome through finish, with Plan and Memory loaded only when useful. Known defects get direct repair, and evidence can change the approach under unchanged acceptance. Fresh exact-content validation remains required. Memory provides optional recall, mining and curation of reviewed topic pages; specialists and the fixed-dispatch adapter remain optional. The change reconciles current documentation and generated skill projections. It preserves native budget and permission authority, BD work ownership, protected external evidence storage and the distinction between a supported lesson and demonstrated later benefit. It adds no scheduler, work store, Go command or evidence schema. Validation: required local Go/build/vet/race checks, aggregate suite, generated-output check and 72 gates pass. The complete 44-test executor suite passes; its shared-deadline fixture now tolerates CI scheduling jitter while still requiring deadline exhaustion and preventing a third launch. Fresh author-distinct review passed all 112 changed paths with no findings; all seven exact-head CI checks passed at bfce33cce. Native restricted-source enforcement and reduced token use are not established by this change.
2026-09-09 10:10:29 -04:00
| RPI traversal | The lean charter: on-demand Plan, Implement/direct repairs and checks, fresh final Validate, then the outcome report. |
Align AgentOps around its operations-layer identity (#1051) Executes docs/plans/2026-08-07-agentops-operations-layer-alignment.md: AgentOps is the operations layer for agentic engineering; the federated integration graph is the topology, the semantic work-and-proof protocol is the contract, and RPI is the standard one-experiment traversal. Retires the ao flywheel command family and all knowledge-flywheel product state, tombstones the seven-move operating-loop workflow, narrows ao init and the .agents state writers to declared destinations, renames the core architecture page to rpi-traversal.md with a compatibility redirect, aligns AGENTS.md, 25 skills, public and package copy, regenerates every owned projection, and strengthens the conformance gates with planted-negative proofs. Both the alignment subject and the follow-up gate-bookkeeping commit carry fresh author-distinct validation PASS verdicts with empty not_checked scope. Test-Removal-Reason: dead knowledge-flywheel and session-store surfaces were deleted with their tests (operations-layer alignment)
2026-08-07 18:37:03 -04:00
The traversal is RPI. The graph is the topology. The protocol defines
interoperability. The operations layer is the product.
`loop` may describe a local control structure inside one tool or a caller's own
campaign. It is not the product category and not the global architecture.
## Supporting terms
| Term | Meaning |
|---|---|
| Context source | A system that owns queryable evidence — the tracker (Beads), Git, CASS, CM, or another caller-selected store. AgentOps cites it with source identity and freshness; it never absorbs its authority. |
| Execution orchestrator | A caller-selected system that schedules, runs, and retries work — a Goal, Mayor, factory, CI, or merge queue. It owns execution lifecycle and never semantic judgment. |
| Software factory | An execution orchestrator with persistent internal sessions and roles (Gas City, NTM rigs, swarms). Operated only through its own doors. |
| Projection | A rebuildable derived view with a named consumer, generator, inputs, exact source identities, and freshness. Deletable without changing semantic behavior; never authority. |
| Fresh judgment | One `PASS | FAIL | NOT_PROVEN` issued over exact content by a context distinct from the candidate author. |
| Runtime completion | A runtime or factory's own report that execution finished. A fact about execution, never validation or delivery proof. |
| Repository check | A deterministic command outcome (`ao gate`, tests, linters). It proves facts; a fresh context judges meaning. |
## Experiment terms
refactor: cut AgentOps to a single-pass evidence loop
2026-07-14 22:01:50 -04:00
| Term | Meaning |
|---|---|
| Intent | The caller's requested behavior and constraints. |
refactor: simplify AgentOps loop and harden CLI
2026-07-15 19:17:08 -04:00
| Intent source | The caller-owned bead, issue, or conversation containing behavior, acceptance, and scope. |
| Intent digest | A runtime-derived digest used to detect acceptance drift; not a model-authored artifact. |
refactor: cut AgentOps to a single-pass evidence loop
2026-07-14 22:01:50 -04:00
| Subject manifest | Deterministic content identity independent of Git. |
| Fresh validator | A declared context identity distinct from the candidate author's identity. |
| Verdict | `PASS`, `FAIL`, or `NOT_PROVEN` over one acceptance digest and one subject digest. |
| RPI report | `NOT_PLANNED`, `NOT_BUILT`, or the semantic verdict, followed by stop. |
Make the engineering harness lean and add optional topic memory (#1116) RPI now owns the authorized outcome through finish, with Plan and Memory loaded only when useful. Known defects get direct repair, and evidence can change the approach under unchanged acceptance. Fresh exact-content validation remains required. Memory provides optional recall, mining and curation of reviewed topic pages; specialists and the fixed-dispatch adapter remain optional. The change reconciles current documentation and generated skill projections. It preserves native budget and permission authority, BD work ownership, protected external evidence storage and the distinction between a supported lesson and demonstrated later benefit. It adds no scheduler, work store, Go command or evidence schema. Validation: required local Go/build/vet/race checks, aggregate suite, generated-output check and 72 gates pass. The complete 44-test executor suite passes; its shared-deadline fixture now tolerates CI scheduling jitter while still requiring deadline exhaustion and preventing a third launch. Fresh author-distinct review passed all 112 changed paths with no findings; all seven exact-head CI checks passed at bfce33cce. Native restricted-source enforcement and reduced token use are not established by this change.
2026-09-09 10:10:29 -04:00
| Revision | A change to accepted outcome/scope requires caller authority; an evidence-driven approach revision within them does not. |
refactor: cut AgentOps to a single-pass evidence loop
2026-07-14 22:01:50 -04:00
| Strategy | Optional advice such as premortem, postmortem, council, or an idea genie. |
| Adapter | Optional transport or runtime that cannot change core semantics. |
Align AgentOps around its operations-layer identity (#1051) Executes docs/plans/2026-08-07-agentops-operations-layer-alignment.md: AgentOps is the operations layer for agentic engineering; the federated integration graph is the topology, the semantic work-and-proof protocol is the contract, and RPI is the standard one-experiment traversal. Retires the ao flywheel command family and all knowledge-flywheel product state, tombstones the seven-move operating-loop workflow, narrows ao init and the .agents state writers to declared destinations, renames the core architecture page to rpi-traversal.md with a compatibility redirect, aligns AGENTS.md, 25 skills, public and package copy, regenerates every owned projection, and strengthens the conformance gates with planted-negative proofs. Both the alignment subject and the follow-up gate-bookkeeping commit carry fresh author-distinct validation PASS verdicts with empty not_checked scope. Test-Removal-Reason: dead knowledge-flywheel and session-store surfaces were deleted with their tests (operations-layer alignment)
2026-08-07 18:37:03 -04:00
## Forbidden conflations
- An **operations layer is not an execution orchestrator**: AgentOps connects
Make the engineering harness lean and add optional topic memory (#1116) RPI now owns the authorized outcome through finish, with Plan and Memory loaded only when useful. Known defects get direct repair, and evidence can change the approach under unchanged acceptance. Fresh exact-content validation remains required. Memory provides optional recall, mining and curation of reviewed topic pages; specialists and the fixed-dispatch adapter remain optional. The change reconciles current documentation and generated skill projections. It preserves native budget and permission authority, BD work ownership, protected external evidence storage and the distinction between a supported lesson and demonstrated later benefit. It adds no scheduler, work store, Go command or evidence schema. Validation: required local Go/build/vet/race checks, aggregate suite, generated-output check and 72 gates pass. The complete 44-test executor suite passes; its shared-deadline fixture now tolerates CI scheduling jitter while still requiring deadline exhaustion and preventing a third launch. Fresh author-distinct review passed all 112 changed paths with no findings; all seven exact-head CI checks passed at bfce33cce. Native restricted-source enforcement and reduced token use are not established by this change.
2026-09-09 10:10:29 -04:00
systems semantically; it does not schedule, own aggregate retry controllers, or own their work.
Align AgentOps around its operations-layer identity (#1051) Executes docs/plans/2026-08-07-agentops-operations-layer-alignment.md: AgentOps is the operations layer for agentic engineering; the federated integration graph is the topology, the semantic work-and-proof protocol is the contract, and RPI is the standard one-experiment traversal. Retires the ao flywheel command family and all knowledge-flywheel product state, tombstones the seven-move operating-loop workflow, narrows ao init and the .agents state writers to declared destinations, renames the core architecture page to rpi-traversal.md with a compatibility redirect, aligns AGENTS.md, 25 skills, public and package copy, regenerates every owned projection, and strengthens the conformance gates with planted-negative proofs. Both the alignment subject and the follow-up gate-bookkeeping commit carry fresh author-distinct validation PASS verdicts with empty not_checked scope. Test-Removal-Reason: dead knowledge-flywheel and session-store surfaces were deleted with their tests (operations-layer alignment)
2026-08-07 18:37:03 -04:00
- An **RPI result is not a factory result**: a traversal's verdict and a
factory's completion report are different facts from different owners.
- **Check success is not semantic PASS**: deterministic green proves facts;
only a fresh validator judges meaning.
- A **projection is not source authority**: derived views cite their sources
and can always be deleted and rebuilt.
refactor: cut AgentOps to a single-pass evidence loop
2026-07-14 22:01:50 -04:00
Avoid using admission, claim, lease, queue, close, land, release, or delivery as
AgentOps lifecycle states. Repositories and callers may use those words in their
own systems; AgentOps does not own those transitions.
Reference in New Issue Copy Permalink