test: 2.5 orchestration run via local qwen3.8-27b (PASS)
check / check (push) Successful in 30s

This commit is contained in:
haylan committed 2026-10-14 14:25:36 +02:00
1 parent 14d6b2ce7a
commit cc9a0ddf6e
3 files changed
+92 -12

No files matched your search

+17 -11
View File
@@ -1,14 +1,16 @@
# Project status # Project status
Updated: 2026-10-14 (Phase 1 COMPLETE — quality gate passed) Updated: 2026-10-14 (2.5 orchestration run PASS — quality gate + end-to-end run both green)
## Current phase ## Current phase
Phase 1 done: plugin installs, loads in OpenCode v2.0.29, skills inject, package builds, Phase 1 done AND 2.5 done: plugin installs, loads in OpenCode v2.0.29, skills inject, package
`@haylan/opencode-pocock@0.1.0` published to the Gitea registry (release run 5684), install-from- builds, `@haylan/opencode-pocock@0.1.0` published to the Gitea registry (release run 5684),
registry verified, nightly workflow proven (dispatch run 5685: github.com reachable, no-change install-from-registry verified, nightly workflow proven (dispatch run 5685: github.com reachable,
early exit). Remaining: Phase 2.5 (manual token-spending orchestration run) and 2.7 (observe a no-change early exit), and a full manual orchestration run of the `pocock` orchestrator against a
week of nightlies). 3-ticket fixture (omniroute/qwen3.8-27b) is **PASS** — all 3 tickets built by parallel workers,
reviewed, fixed, merged, clean exit, no product bugs. Remaining: 2.7 (observe a week of
nightlies).
## Done ## Done
@@ -41,8 +43,12 @@ week of nightlies).
## Known gaps / risks ## Known gaps / risks
- 2.5 not done: no token-spending end-to-end orchestration run yet (pocock building a fixture - 2.5 DONE (2026-10-14): full orchestration run PASS on omniroute/qwen3.8-27b — see
spec with parallel workers). Drift/unit/smoke suites cover the deterministic surface. docs/runtime-checks.md § "2.5 orchestration run". Noted env-only quirks: skill id collision
resolves `code-review` to the user's global `~/.agents/skills` copy (documented, not a plugin
bug); user's codebase-memory "graph-first" rule rejected one orchestrator `glob` (recovered via
`find`). Local 27b took ~63 min; reviewer subagents stream output only on completion (looked
stalled, was not). No product bugs found.
- 2.7: nightlies need a week of green runs before "stable" is declared; first scheduled run - 2.7: nightlies need a week of green runs before "stable" is declared; first scheduled run
2026-10-15 03:30 UTC. 2026-10-15 03:30 UTC.
- Background subagents in V2 (env-gated in V1) not verified; orchestrator text works either way. - Background subagents in V2 (env-gated in V1) not verified; orchestrator text works either way.
@@ -53,10 +59,10 @@ week of nightlies).
## Next steps (exact) ## Next steps (exact)
1. 2.5: manual orchestration run — fixture project with spec + 3 tickets (2 independent, 1 1. ~~2.5: manual orchestration run~~ DONE 2026-10-14 (PASS, see docs/runtime-checks.md).
blocked), run `opencode --agent pocock`, verify worktree loop/merges/review/permissions;
record in docs/runtime-checks.md.
2. Watch nightly runs for a week (2.7); on upstream release (17 pending changesets), verify the 2. Watch nightly runs for a week (2.7); on upstream release (17 pending changesets), verify the
full sync→generate→publish→commit-back cycle end-to-end. full sync→generate→publish→commit-back cycle end-to-end.
3. Then declare stable and announce; keep classification.json current when upstream promotes 3. Then declare stable and announce; keep classification.json current when upstream promotes
in-progress skills. in-progress skills.
4. (Optional, owner) Decide whether to fold the 2.5 agent-improvement suggestions (truncated-skill
note, non-ff resume reminder) into agents/*.md — they are in the runtime-checks.md report.
+8 -1
View File
@@ -48,8 +48,10 @@ State: `.opencode/state/status.md`. Design: `docs/architecture.md`.
permissions (edit/subagent/webfetch deny, git-read shell allow-list, skill allow code-review) permissions (edit/subagent/webfetch deny, git-read shell allow-list, skill allow code-review)
- [x] 2.4 orchestrator V2 flow wiring — done in 1.6: agent text covers background AND batch - [x] 2.4 orchestrator V2 flow wiring — done in 1.6: agent text covers background AND batch
dispatch; slash UX + /tmp default documented in README from runtime findings dispatch; slash UX + /tmp default documented in README from runtime findings
- [ ] 2.5 per-skill regression checks (drift suite exists; add manual token-spending orchestration - [x] 2.5 per-skill regression checks (drift suite exists; add manual token-spending orchestration
run: fixture spec + 3 tickets like pocock-agents smoke) + record in docs/runtime-checks.md run: fixture spec + 3 tickets like pocock-agents smoke) + record in docs/runtime-checks.md
(DONE 2026-10-14: full PASS on omniroute/qwen3.8-27b, ~63 min, all 3 tickets + review + fixes,
no product bugs — see docs/runtime-checks.md § "2.5 orchestration run")
- [x] 2.6 README usage guide + finalize skill-mapping (matrix generated from classification.json) - [x] 2.6 README usage guide + finalize skill-mapping (matrix generated from classification.json)
- [ ] 2.7 one week of green nightlies -> declare stable (BLOCKED on PACKAGE_TOKEN secret) - [ ] 2.7 one week of green nightlies -> declare stable (BLOCKED on PACKAGE_TOKEN secret)
@@ -69,3 +71,8 @@ State: `.opencode/state/status.md`. Design: `docs/architecture.md`.
## Done log ## Done log
- 2026-10-14: phase-0 analysis docs written (AGENTS.md, state/, docs/architecture.md, TODO.md). - 2026-10-14: phase-0 analysis docs written (AGENTS.md, state/, docs/architecture.md, TODO.md).
- 2026-10-14: 2.5 manual orchestration run — `pocock` orchestrator drove a 3-ticket `greeting`
fixture (omniroute/qwen3.8-27b, ~63 min) end-to-end: parallel workers on isolated worktrees,
non-ff resume path exercised, two-axis code review, review-fix worker, all 16 tests green on
`main`, worktrees cleaned, tickets marked done, `EXIT_CODE=0`. **PASS, no product bugs.**
Recorded in docs/runtime-checks.md § "2.5 orchestration run".
+67
View File
@@ -55,3 +55,70 @@ password printed on stdout) in a temp fixture and asserts against the HTTP API.
harmless, retried on next run. harmless, retried on next run.
- The user's global `~/.agents/skills` already contains most Matt Pocock skills — smoke assertions - The user's global `~/.agents/skills` already contains most Matt Pocock skills — smoke assertions
are written to tolerate id collisions (presence + flags, not source path). are written to tolerate id collisions (presence + flags, not source path).
## 2.5 orchestration run (2026-10-14, omniroute/qwen3.8-27b)
**Result: PASS** — full `implement-spec` flow, all 3 tickets, review, fixes, clean exit.
Fixture: fresh clone of the pocock-agents `greeting` reference (AGENTS.md `## Agent skills` block →
`docs/agents/issue-tracker.md` local tracker, GLOSSARY.md, `.scratch/greeting/spec.md` + 3 tickets —
01+02 independent, 03 blocked by 01+02, `src/greeting.js` + node:test, no remote, no deps). Agents
copied to `.opencode/agents/`; `opencode.json` =
`{"$schema":…,"plugins":["G:/_DEV/repos/mattpocock-skills-opencode/dist"],"model":"omniroute/qwen3.8-27b"}`
(`model` is a valid top-level v2 config key — confirmed against `opencode.ai/config.json`).
Command (cwd = fixture, background, `> run.log 2>&1`):
```
opencode run --agent pocock --auto "Implement the spec in .scratch/greeting/spec.md using
implement-spec: build all three tickets with parallel workers, review, and finish. The tracker is
local markdown (see docs/agents/issue-tracker.md). Do not ask me anything else; proceed through the
whole flow autonomously."
```
Duration: ~63 min (11:00:26Z → 12:03:47Z) on the local 27b model. `EXIT_CODE=0`.
| # | Check | Result | Evidence |
| --- | -------------------------------------------------- | ------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| 1 | Integration + ticket branches exist and are merged | PASS | `spec/greeting` is an ancestor of `main`; `git merge-base --is-ancestor` confirms 01, 02, 03 all in `spec/greeting`. 01 + 03 ff-merged; 02 diverged after 01 landed → resolved via merge commit `0308c30 "Merge branch 'spec/greeting' into ticket/greeting/02"`, exactly the agent's documented resume-worker path. |
| 2 | Worktree dir used + cleaned | PASS | All worktrees under `/tmp/pocock-workers/pocock-25-fixture/greeting/` (`$WT` default). `git worktree list` shows only the main checkout after the run; orchestrator logged `git worktree remove --force … tickets/{01,02,03,review-fixes}` → `CLEANED`. Leftover empty `notes/` + `tickets/` dirs (no worktree admin entries) — harmless. |
| 3 | Tests pass on integration branch | PASS | Fresh worktree of `spec/greeting`: `npm test` = 16/16 pass, 0 fail (also 16/16 at every in-run merge point). |
| 4 | Flow evidence in run.log | PASS | `→ Skill "ask-matt"`, `→ Skill "implement-spec"`, `→ Skill "code-review"`; 4 `pocock-worker` dispatches (ticket 01/02 parallel, "Worker 02: merge integration tip" resume, ticket 03) + 4 `pocock-reviewer` dispatches (Spec axis, Standards axis, + compact re-runs); TDD red→green commit pairs per ticket (`test(…): red…` → `feat(…)`); no `✗` permission denials — zero worker escape attempts (no push/stash/out-of-allow-list skill). Only failure in the whole log: one `✗ Glob` hit by the user's global codebase-memory "graph-first" rule (environment quirk, below). |
| 5 | Model actually served | PASS | Every turn banner: `> pocock · qwen3.8-27b` — top-level `"model"` in opencode.json took effect for orchestrator and all nested subagents. |
| 6 | Ticket statuses updated | PASS | All three `issues/NN-*.md` now read `Status: done` with a `## Comments` note naming the branch — per the local-tracker convention in `docs/agents/issue-tracker.md`. |
Flow detail worth keeping: 02's divergence was handled by resuming worker 02 to merge the new
integration tip (agent text: "It knows its own change best, so it's the merger") — the non-ff path
worked, not just the happy ff path. Review produced 2 actionable Standards findings + 1 real Spec gap
(farewell error-message criterion untested); orchestrator dispatched a `review-fixes` worker for them
and merged back to `main`. `spec/greeting` then ff-merged to `main` (no remote → delivered on
default branch, per agent text).
Product bugs: **none**. Everything above is agent/plugin behaving per its documented design.
Findings / model-weakness notes (not product bugs):
- **Skill-id collision resolved to the user's global copy.** When the orchestrator loaded
`code-review`, the content came from `~/.agents/skills/code-review/SKILL.md` (pre-existing global
Matt Pocock install), not the plugin-injected copy — same collision documented in the 1.10 note.
The model then said the skill text was "truncated" and hunted for the SKILL.md on disk (found the
global one). Our plugin ships the full body (`code-review` body = 6177 chars, no slicing in
`src/index.ts`), so this is the collision + local-model behaviour, not truncation by the plugin.
- **`graph-first` rule collision.** The user's global codebase-memory-mcp injects a rule that forces a
graph query before any `glob`. One orchestrator `glob` was rejected with that error; the model
recovered by using `find`. Not our product; note for anyone running this with codebase-memory
installed.
- **Local 27b is slower than capable.** The whole run took ~63 min; reviewer subagents were CPU-busy
for ~14 min with no log growth (subagent output streams on completion, not incrementally — expected,
not a stall). No stalls requiring intervention; no permission denials; the only judgment shortcuts
were self-declared in the final summary (skipped a second post-fix review pass; left
spec-prescribed module duplication alone).
Suggested agent improvements (NOT applied — for the owner):
- `pocock.md`: consider noting that when a skill's content looks truncated, the model may re-read the
SKILL.md from disk; if the user also has a global Matt Pocock install, the id resolves to the file
copy — worth a one-line "skills may resolve to a global copy; content is authoritative either way".
- `pocock-worker.md`/`pocock.md`: the non-ff resume path works but is model-dependent; a short
reminder that "a ticket that can't ff-merge is normal when a parallel ticket merged first — resume
its worker" could make the 27b class of model take the documented path more reliably.