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

Files

T

boshu 4a7c6a0701 feat(gates): docs.skill-refs --all-docs — widen the 6-file skill-ref gate to live docs + SKILL-ROUTER sync (age-gate-the-ungated-egwt.5)

Add an --all-docs mode to check-doc-skill-refs.sh that scans the union of the
pinned doctrine files PLUS the full LIVE docs/** set (scripts/lib/docs-scope.sh),
ratcheted by a new filename-pinned shrink baseline
(scripts/.docs-skill-refs-baseline). Two-way ratchet: a non-baselined live doc
with a dead `/skill` ref FAILS; a baselined file that no longer offends FAILS
demanding a prune. Detection stays slash-syntax + headings only (never bare
skill names). Default (no-flag) behavior is byte-identical.

docs/SKILL-ROUTER.md (the curated router) is covered by --all-docs (a live
docs/** file, never baselined) so a dead ref there FAILS strict, without
perturbing the byte-identical default doc count.

seed.go docs.skill-refs row: Match widened to docs/** + the baseline + the lib;
Args set to [--all-docs --strict]. Burn-down (long-tail only): repointed retired
`/retro`->/post-mortem, `/vibe`->/validate, `/ratchet`->/flywheel,
`/research-multi`->/research, `/implement-wave`->/crank, and reworded the
retired `/dream` across 13 docs/levels + docs/workflows + docs/philosophy
pages, then pruned them from the baseline (91 -> 78 offenders). Owner-locked
pages (comparisons/getting-started/positioning) stay baselined.

Bats twin tests/scripts/check-doc-skill-refs-all-docs.bats covers all the modes.

2026-07-01 18:12:53 -04:00

2.4 KiB

Raw Blame History

description

description
Execute an epic to completion using swarm for parallel wave execution

/crank

Runs an entire epic through waves of parallel execution until ALL children are CLOSED.

Usage

/crank <epic-id>
/crank ao-epic-123

Architecture

Crank is the autonomous orchestrator that uses swarm for each wave:

Crank (orchestrator)           Swarm (executor)
    |                              |
    +-> br ready (wave issues)     |
    |                              |
    +-> TaskCreate from beads  --->+-> Spawn agents (fresh context)
    |                              |
    +-> /swarm                 --->+-> Execute in parallel
    |                              |
    +-> Verify + br update     <---+-> Results
    |                              |
    +-> Loop until epic DONE       |

Separation of concerns:

Crank = Beads-aware orchestration, epic lifecycle, knowledge flywheel
Swarm = Fresh-context parallel execution (Ralph Wiggum pattern)

How It Works

The FIRE Loop:

FIND: BEADS_DIR="$(ao beads dir)" br ready - get unblocked beads issues
IGNITE: Create TaskList tasks, invoke /swarm
REAP: Swarm collects results, crank syncs to beads
ESCALATE: Fix blockers, retry failures
Loop until all children are CLOSED

Output

/crank ao-epic-123

Epic: "Add user dashboard"
Total: 8 issues

[Wave 1] br ready → [ao-1, ao-2, ao-3]
         TaskCreate for each
         /swarm → 3 agents spawned
         ao-1 DONE, ao-2 DONE, ao-3 BLOCKED

[Wave 2] br ready → [ao-4, ao-5, ao-3]
         TaskCreate for each
         /swarm → 3 agents spawned
         ...

[Final Vibe] Running /vibe on recent changes...
             All checks passed.

<promise>DONE</promise>
Epic: ao-epic-123
Issues completed: 8
Waves: 4/50

Limits

MAX_EPIC_WAVES = 50 - Prevents infinite loops
Swarm handles parallelism per wave (no max agent limit in swarm)

Failure Handling

Crank handles failures automatically:

Retry failed issues in next wave
Skip blocked issues (revisit when unblocked)
Escalate persistent failures after 3 retries

When to Use

Scenario	Skill
Execute entire epic autonomously	`/crank`
Just parallel execution (no beads)	`/swarm` directly
Single issue	`/implement`

/validate - Runs automatically at end
/post-mortem - Extract learnings after epic completes

2.4 KiB Raw Blame History