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>
2.3 KiB
2.3 KiB
name, description, category, interval, stop-condition, components, tags
| name | description | category | interval | stop-condition | components | tags | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| goal-refiner-loop | Rewrites a vague request into a precise goal — exact end state, how to verify it, what not to touch, and the stop condition — then executes against it. | engineering | on-demand | The refined goal's verifiable end state is reached, or its explicit turn/budget cap is hit. |
|
|
Goal-Refiner Loop
Loop Engineering — "your agent is not dumb, your instructions are just vague." This loop's only first job is to turn a fuzzy ask into a rigorous, verifiable goal before any code is written, then run the
/goalagainst it.
🎯 Goal
Convert a vague request into a precise goal — the exact end state, how it will be verified, what must not be touched, and the stop condition — confirm it, then execute.
⏱️ Schedule
Trigger: on-demand.
▶️ Run it
/goal Before doing anything, rewrite my request into a precise goal: the exact end state, how you will verify it, what you must not touch, and the stop condition (including a turn cap). Confirm that goal with me, then execute against it and stop when the end state is verified.
🔁 Iteration steps
- Refine —
critical-thinkingrewrites the ask into a crisp, testable goal. - Scope —
code-architectdefines the blast radius: what's in, what must not be touched. - Confirm — surface the refined goal for a quick yes before acting.
- Act — execute against the goal.
- Verify — check the end state against the stated verification; stop when met.
🛑 Stopping condition
The goal's verifiable end state is reached, or the turn/budget cap is hit (the loop reports how far it got).
💰 Budget & guardrails
Every refined goal carries an explicit "what not to touch" and a turn cap, so an under-specified ask can't sprawl into a risky, open-ended run.
🧩 Referenced components
agent:expert-advisors/critical-thinking— sharpens the goal and verification.agent:development-team/code-architect— bounds scope and the do-not-touch list.
💡 Example
"Make the app faster" becomes "p95 of /search ≤ 200ms measured by the existing benchmark, without changing the public API, stop after 8 turns" — and the loop executes against that.