agentleFS
Sign inStart free
maidepsmaideps/ai-harness-framework/skills/feature-cycle/SKILL.md

feature-cycle

Skill2 starsChanged 6 days agoMITView on GitHub

The one-feature-at-a-time loop — activate, implement, verify through the gate, record evidence.

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 2

    1. Tetteh-Yaw-Precious/feature-cycle★ 2same name
    2. code-with-mehedi/nextjs-claude-starter★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository11 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read feature_list.json — it is the source of truth, not chat history.

    line 17
  2. Very specific:

    Run npm run check before every commit.

    line 22
  3. Very specific:

    When the behavior is implemented, commit the code and its docs (after npm run check), then run the gate:

    line 23
  4. Distinctive:

    Update docs in the same commit as the code change — no stale documentation.

    line 31

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Feature Cycle

The core workflow of this harness. Everything else serves this loop.

When to Use

  • Starting any implementation work on a feature
  • Picking up a partially done feature in a new session
  • Deciding what "done" means before writing code

Workflow

  1. Read feature_list.json — it is the source of truth, not chat history. To add a feature, use npm run add-feature -- <id> <name> [--behavior ..] [--verification ..] [--depends a,b] [--layer label=cmd] instead of hand-writing the JSON.
  2. Pick exactly ONE unfinished feature. Leave every other feature alone.
  3. Activate it with npm run activate -- <id>; the command refuses a second active feature or unmet dependencies and writes the state shard.
  4. Confirm the contract: at most one active feature (WIP=1), and every dependency of the active feature must already be passing. npm run test enforces both.
  5. Implement only that feature. Stay in scope — do not touch files unrelated to it. The feature's behavior is the spec: if an existing test contradicts it, the test is wrong — fix it in the same commit and say so. Stop and ask only if the behavior itself looks wrong.
  6. Run npm run check before every commit. A layer that reports SKIP does not count as verified; a FAIL stops the chain.
  7. When the behavior is implemented, commit the code and its docs (after npm run check), then run the gate:
    npm run verify-feature -- <feature-id>
    The gate runs the feature's own verification layers, refuses to pass it while dependencies are unmet, and records evidence in .harness/state/<feature-id>.json — feature_list.json itself is never rewritten. Never mark passing by hand. The evidence names a commit, so the gate refuses while code is uncommitted. Harness bookkeeping (.harness/state/, progress/, PROGRESS.md, …) doesn't count; commit the shard it writes, with the session notes, in the next commit. Use --allow-dirty only when a human agrees: the evidence is then marked bound: false.
  8. The gate records the trail itself (.harness/trails/<stamp>-verify-feature-<id>.json) on a bound pass; commit it with the shard. npm run vcr is only for a full-run record outside a feature.
  9. Name the feature in docs/quality-document.md and rate the modules it touched: npm run clean-check fails until you do.
  10. Update docs in the same commit as the code change — no stale documentation.

Quick Reference

  • Exactly one feature active before starting
  • Dependencies of the active feature are passing
  • npm run check before each commit
  • Code and docs committed, then npm run verify-feature -- <id>
  • Evidence recorded in .harness/state/<id>.json, trail in .harness/trails/
  • Feature named and its modules rated in docs/quality-document.md; npm run clean-check passes
  • Docs updated in the same commit

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free