mirror of https://github.com/boshu2/agentops.git synced 2026-09-14 15:08:13 +08:00

Files

T

Bo 7a765cde19 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

2.4 KiB

Raw Blame History

Goals

Product outcome

AgentOps is the operations layer for agentic engineering: it makes one coding-agent experiment independently inspectable without taking over the consumer's engineering system.

The canonical outcome is:

explicit behavior
  -> bounded implementation experiment
  -> exact content identity
  -> fresh independent judgment
  -> durable PASS | FAIL | NOT_PROVEN

Fitness properties

Behavior before activity. The caller-owned intent states the active behavior, acceptance examples where useful, non-goals, evidence, and bounded write scope before implementation begins.
One experiment. Implement performs one RED -> GREEN -> refactor cycle and reports facts without retry or delivery authority.
Fresh judgment. PASS requires explicit, distinct author and validator context IDs plus a freshness attestation.
Exact subject. Content identity covers files, symlinks, deletions, executable bits, roots, exclusions, and canonical digests without requiring Git or ao.
Honest uncertainty. Mutation, incomplete changed-path coverage, missing identity, or missing proof returns NOT_PROVEN. Proven scope or acceptance failure returns FAIL.
Sovereign proof. Validate atomically writes content-addressed JSON to caller-controlled storage. Provenance is optional audit.
Stop boundary. RPI dispatches Plan, Implement, and Validate at most once, reports the result, and emits no next action.
Open ecosystem. Callers keep their trackers, Git, PRs, CI, cloud agents, merge queues, rollback, and release systems.

Structural constraints

Core hard dependencies are only rpi -> {plan, implement, validate}.
Learn and all strategy/factory/specialist skills are off-path or optional.
Core schemas contain no retry, budget, queue, claim, lease, admission, next-action, closure, release, or delivery state.
The pure manifest and verdict helpers make no Git, tracker, queue, network, release, or delivery call.
Deterministic ao gate check reports repository check success or failure only.
Semantic validation is a skill responsibility, not a CLI state machine.

Measured learning hypothesis

Collections of durable verdicts may reveal repeated defect classes that deserve better context, tests, or checks. Learn may propose evidence-backed candidates off-path. Promotion requires separate evaluation; no observation silently becomes policy or changes a prior verdict.