mirror of
https://github.com/davila7/claude-code-templates.git
synced 2026-09-19 01:30:23 +08:00
Daniel Avila
7f890b1b3b
* feat: add Loops component type with referenced components Introduce "loops" as a new first-class component type — autonomous agentic workflows (goal + interval + stop condition) inspired by Loop Engineering and the Forward Future Loop Library. A loop references other components (agents, skills, hooks, commands) and the dashboard renders those references as clickable cards on the loop detail page. - Add 10 starter loops under cli-tool/components/loops/ (engineering, evaluation, operations) - Catalog generator: scan loops + write per-component content files - Dashboard: new type in icons/types/data, /loops page, and a "Referenced Components" section on the loop detail page - CLI: --loop flag installs the loop to .claude/loops/ and auto-installs its referenced components - Docs: CLAUDE.md and cli-tool rule updated Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01T9ynMiCHt8JwdMd8KjCkES * feat: add 8 more loops (goal/loop/schedule patterns) Expand the loops catalog to 18, drawing ideas from community "Loop Engineering" patterns. Adds loops across the three trigger verbs (/goal, /loop, /schedule) and emphasizes the verifier-in-the-loop and budget / anti-spin guardrails: - engineering/five-minute-maintainer-loop - engineering/build-test-fix-loop - engineering/anti-spin-build-loop - engineering/completion-contract-loop (goal) - engineering/goal-refiner-loop (goal) - engineering/adversarial-review-loop (cross-model) - operations/overnight-pr-routine-loop (schedule) - evaluation/human-approval-loop (human-in-the-loop) All referenced components verified to exist; catalog regenerated. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01T9ynMiCHt8JwdMd8KjCkES --------- Co-authored-by: Claude <noreply@anthropic.com>
2.6 KiB
2.6 KiB
name, description, category, interval, stop-condition, components, tags
| name | description | category | interval | stop-condition | components | tags | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| anti-spin-build-loop | Builds toward a machine-checkable contract with explicit anti-spin guardrails — stops on no progress, repeated approaches, flip-flopping, or a spend budget. | engineering | 15m | The machine-checkable contract passes, or an anti-spin guard trips (no progress, repeated/flip-flopping approach, or budget hit). |
|
|
Anti-Spin Build Loop
Loop Engineering — most agent loops never stop to ask whether they're actually making progress, so they retry the same broken approach or quietly edit the test to pass. This loop runs build → audit → verify against a machine-checkable contract, with explicit stops that detect spinning.
🎯 Goal
Drive toward a verifiable contract while actively guarding against the classic failure mode: an agent that loops forever without converging.
⏱️ Schedule
Suggested interval: 15m.
▶️ Run it
/loop 15m "Build toward the goal, then audit and verify against a machine-checkable contract. Track progress each pass. Stop if you make no progress, repeat an approach you already tried, flip-flop between two approaches, or hit the budget. Finish only when the contract passes — never by weakening the contract or the tests."
🔁 Iteration steps
- Build — advance toward the contract.
- Audit —
architect-reviewchecks the change;scope-guardblocks weakening the tests/contract. - Verify — run the machine-checkable contract; record progress vs. last pass.
- Anti-spin check — no progress? repeated approach? flip-flop? budget hit? → stop and report.
- Observe — if making progress and contract not yet met, loop; else
/pr-reviewand hand off.
🛑 Stopping condition
Contract passes, or an anti-spin guard trips — whichever comes first.
💰 Budget & guardrails
Hard caps on iterations and spend; no-progress, repeated-approach and flip-flop detection. The contract and tests are immutable — the loop may not edit them to "pass".
🧩 Referenced components
agent:expert-advisors/architect-review— independent audit each pass.command:git-workflow/pr-review— structured review at handoff.hook:quality-gates/scope-guard— prevents weakening tests/contract to fake success.
💡 Example
Chasing a contract for a parser, the loop tries approach A, then B, then drifts back to A — the flip-flop guard trips and it stops with a clear report instead of burning tokens forever.