agentleFS
lgzarturolgzarturo/codeconductor/.agents/skills/openspec/SKILL.md

openspec

Skill0 starsChanged 26 days agoMITView on GitHub

This skill delivers an existing BACKLOG.md item. It is a workflow with CLI gates, not a reference doc. Specs describe WHAT; design.md describes HOW. NOT for creating BACKLOG.md (use skill backlog) or for stack-specific coding rules. The CLI is always npx cc-codeconductor. 1. openspec validate — must pass before delivery. If you reached this workflow on your own (the user did not ask for OpenSpec) and there is no BACKLOG.md / openspec/ root, answer normally instead — never scaffold one…

7 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When implementation reveals a design problem, pause and reconcile the planning artifacts first — in any direction (a later artifact may …

    line 50
  2. Very specific:

    When the task concerns web layout, component states, feedback, motion, or a requested UI audit, apply skill web-design-engineering from the …

    line 72
  3. Very specific:

    When an item concerns concrete Jetpack Compose phone layout, component states, interaction, accessibility, visual behavior, or a requested …

    line 83

“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “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

OpenSpec delivery

Overview

This skill delivers an existing BACKLOG.md item. It is a workflow with CLI gates, not a reference doc. Specs describe WHAT; design.md describes HOW.

When to Use

  • /cc-openspec or openspec plan / next / done / sync / verify / archive
  • An item is READY or later and must move through the state machine

NOT for creating BACKLOG.md (use skill backlog) or for stack-specific coding rules.

Process

The CLI is always npx cc-codeconductor.

  1. openspec validate — must pass before delivery. If you reached this workflow on your own (the user did not ask for OpenSpec) and there is no BACKLOG.md / openspec/ root, answer normally instead — never scaffold one as a side effect.
  2. openspec plan BC-xxx if the item is not yet PLANNED. Planning only: it writes TaskCards plus proposal/design/tasks/specs under openspec/changes/<slug>/ and stops. Never implement in this step.
  3. Review the plan before start: read proposal → delta specs → tasks, in that order, and confirm intent, scope, testable FR/SC, and edge-case scenarios. Fix the Markdown directly or ask for revisions — code comes later.
  4. openspec analyze --output json — CRITICAL findings exit 1. Do not implement.
  5. Phases: discover (repo-explorer) → design (architect) → test (tester) → implement (implementer) → review (reviewer). If Global TDD required: yes, test runs before implement. Discover is read-only: it never writes code.
  6. openspec done on test/implement needs runner evidence: run npx cc-codeconductor tdd capture --task <cardId> --phase red|green --command "<test command>" first (RED on the test card, GREEN on the implement card). Handmade evidence JSON is rejected. It is ev-tdd-* evidence, not an RDD receipt (ev-rdd-*, rdd verify). done ticks that card's tasks.md boxes and keeps your edits; the implementer ticks any other box it adds (- [ ] → - [x], only x/X counts). archive fails while boxes stay unchecked (override: --allow-unchecked).
  7. When implementation reveals a design problem, pause and reconcile the planning artifacts first — in any direction (a later artifact may force revising an earlier one). Planning artifacts only in that step, never code; confirm each edit. If the item's intent changed rather than its details, open a fresh item with /cc-backlog (/cc:backlog) instead of warping this one.
  8. openspec verify --output json — advisory pre-archive checklist (archiveReady plus Completeness/Correctness/Coherence issues). Optional: openspec sync merges delta specs into openspec/specs/ without closing.
  9. scorecard create --task BC-xxx --from-diff then record a verdict.
  10. openspec archive only after human review when Review required: yes and the scorecard is PASS. Archive re-checks planning artifacts and analyze CRITICALs, and fails on unchecked tasks.md boxes (override: --allow-unchecked), ticks the BACKLOG acceptance criteria.

Status machine: TODO → READY → PLANNED → IN_PROGRESS → REVIEW → DONE → Archive. BLOCKED returns to READY. Reviewer rejection: REVIEW → IN_PROGRESS.

openspec status --output json also reports artifact presence, tasks.md checkbox progress, and nextSteps for the next CLI call.

Web interface scope

When the task concerns web layout, component states, feedback, motion, or a requested UI audit, apply skill web-design-engineering from the installed skill library. Use it during intake/discovery, design, test/implement, and review as applicable; keep this workflow’s gates and TDD order. Framework presence alone does not activate it; backend-only and native mobile work are excluded. Record required visual checks as pending when unavailable. Keep findings in the existing Review Report: contract failures in Spec Axis, technical issues in existing subchecks, and aesthetic preferences as suggestions; do not add an axis.

Android phone interface scope

When an item concerns concrete Jetpack Compose phone layout, component states, interaction, accessibility, visual behavior, or a requested UI audit, apply skill android-ui-design throughout discover, design, test/implement, and review. Keep OpenSpec gates and TDD order. Kotlin, Compose, framework, or dependency presence alone does not activate it; non-UI Android work stays with android. Keep audit-only delivery read-only and record unavailable visual or emulator evidence as pending.

Common Rationalizations

RationalizationReality
Validate is bureaucracyopenspec validate is the gate. Skipping it is a defect.
I'll add tests after greenGlobal TDD required means tester before implementer.
I'll write the evidence JSON myselfHandmade TDD JSON is rejected. Use the verification runner.
The item is small; skip analyzeopenspec analyze CRITICAL still stops implement.
The plan is generated; skip reading itRead proposal → specs → tasks before start. Generated is not reviewed.
I'll fix the spec after shippingReconcile planning artifacts before continuing to implement.

Red Flags

  • Implementing while analyze reports CRITICAL
  • Starting cards before reading the generated plan
  • Archive without a PASS scorecard when review is required
  • Archive while openspec verify reports CRITICAL
  • Acceptance like "improve UX" with no measurable check

Verification

  • openspec validate exit 0
  • Plan reviewed (proposal → specs → tasks) before start
  • openspec analyze --output json has no CRITICAL
  • TDD evidence from the runner when TDD is required
  • tasks.md boxes ticked as FRs land
  • openspec verify --output json checked before archive
  • scorecard create --from-diff recorded
  • Suite check (optional): npx cc-codeconductor scorecard suite-run --suite workflow-gates

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.