team-feature
Skill4 starsChanged 46 days agoMITView on GitHub
Delivers one feature end-to-end as a vertical slice: feature-spec check → branch → API contract → data model → backend → frontend/game → tests → security review → commit and PR → code review, spawning the right leads and engineers in parallel where independent. Use for 'build feature F-NNN', 'implement the whole feature', 'deliver this end-to-end'.
- Commits and pushes
1 other project uses a version of this.
Adopted by
Most recently taken up by
- tranhieutt/software_development_departmenttheir own version, first seen Apr 30, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- tranhieutt/software_development_department★ 71same 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 repository52 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- testing/CLAUDE.mdCLAUDE.md
- a11y-auditSkill
- adoptSkill
- api-contractSkill
- architecture-decisionSkill
- architecture-reviewSkill
- backlogSkill
- brainstormSkill
- changelogSkill
- code-reviewSkill
- collaboration-protocolSkill
- create-storiesSkill
- data-modelSkill
- dependency-auditSkill
- deploySkill
- design-systemSkill
- dev-storySkill
- docsSkill
- feature-specSkill
- game-conceptSkill
- hardenSkill
- helpSkill
- hotfixSkill
- impactSkill
- incidentSkill
- initSkill
- migrateSkill
- pentestSkill
- perf-auditSkill
- product-specSkill
- qa-planSkill
- refactorSkill
- release-checklistSkill
- retrospectiveSkill
- security-auditSkill
- setup-stackSkill
- skill-improveSkill
- skill-testSkill
- sprint-planSkill
- sprint-statusSkill
- stack-updateSkill
- startSkill
- story-doneSkill
- team-gameSkill
- team-releaseSkill
- team-securitySkill
- tech-debtSkill
- test-setupSkill
- threat-modelSkill
- updateSkill
- ux-specSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
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
- Very specific: line 49
Stage by name: Read git status --short first and deal with every unplanned ??
- Very specific: line 52
Wait for CI: only for a run that exists: one background gh run watch <run-id> --exit-status.
≈ “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.
Files
Team: Feature
Reply in the project conversation language (CLAUDE.md → Language); code, identifiers, paths and commit messages stay in English.
Orchestration. File writes and any mutation (git, deploy) happen only after an explicit "May I write?" / "Proceed?" — each one AskUserQuestion (proceed (Recommended) · show the draft/diff first · not now) → "yes". After the "write" answer: touch .claude/.write-consent (rule 7 — the consent-guard hook checks the marker). A partial report on BLOCKED at any stage is mandatory.
Consent for delegated work is collected by this skill. A subagent spawned through Task cannot ask the user. Before each Task batch, the "Proceed?" question names the agents and the files each will create or change; that answer covers those files. A specialist that needs to go beyond its brief (another file, a new dependency, a contract or schema change) stops and reports it, and the parent asks.
Agents below are spawned with a studio subagent_type: web-studio:<name> in plugin mode, <name> in copy mode. Commands are /web-studio:<name> in plugin mode and /<name> in copy mode; a bare /code-review in plugin mode runs Claude Code's built-in review, not the studio's (coordination-rules § Subagents). This skill runs another studio skill through the Skill tool, one skill after another; the called skill keeps all of its phases and gates, and its verdict comes back as text this skill quotes. ‖ below means one Task batch of agents launched together — never two skills at once. In the commands below, <hooks> is .claude/hooks/ in copy mode and ${CLAUDE_PLUGIN_ROOT}/hooks/ in plugin mode.
Phase 1: Readiness
- A feature spec with criteria — missing → run
/feature-specthrough theSkilltool and read its verdict line: continue only onAPPROVED;NEEDS REVISIONorBLOCKED→ stop withBLOCKED (stage 1: feature-spec <verdict>)and the partial report (the spec's open points, nothing else started). Technical-preferences;.claude/docs/git-workflow.md. - Review mode:
--reviewwhen given, elseproduction/review-mode.txt(leanwhen neither exists). The mode scopes reviews only (review-workflow.md § Review mode) — the "Proceed?", "May I write?" and commit gates are asked in every mode:full— Phase 5 runs every check:appsec-engineer‖accessibility-specialist(UI) ‖performance-engineer, all three.lean—appsec-engineeronly when the slice touches auth, data or the network;accessibility-specialistwhen there is UI;performance-engineerwhen a budget is at risk.solo— no review gates: Phase 5 runs only what the user picks in oneAskUserQuestion(appsec-engineer(Recommended when the slice is security-sensitive) ·accessibility-specialist·performance-engineer· none); nothing blocks the slice.appsec-engineerstays mandatory for security-sensitive paths in every mode (rule 6).
- Scope: the stories the slice delivers (
production/stories/F-NNN/). One story → the branch and the commit scope areS-NNN; several →F-NNN.
Phase 2: Branch
Per .claude/docs/git-workflow.md (one story = one branch = one PR), as /dev-story Phase 3:
- One
AskUserQuestion: branchfeat/<scope>-<slug>from an up-to-date default branch and start (Recommended) · stay on the current branch<name>(it already belongs to this slice) · stop. The "start" answer is the consent for the branch and the session-state update. git fetch origin. If the current branch is the default branch, or is already merged intoorigin/<default>(git merge-base --is-ancestor HEAD origin/<default>):git switch <default> && git pull --ff-only origin <default>. Never continue on a merged branch.git switch -c feat/<scope>-<slug>.- Record the start through the studio's writer:
<hooks>session-state.sh set Task "F-NNN …" Branch feat/<scope>-<slug> Next "/code-review --diff". - Stamp the story cards (as
/dev-storyPhase 3 step 5): on the metadata line of every story card the slice implements (production/stories/F-NNN/S-NNN-*.md), setStatus: In Progressand writeStarted: YYYY-MM-DDTHH:MMwith the actual time —/story-donemeasures the story's actual duration from it and refuses a card that never leftReady. The "start" answer of step 1 covers these edits; the cards ride the slice'sfeat(<scope>)commit in Phase 6 (staged by name), so the stamp never lands on the default branch by itself.
Phase 3: Contract and data (in parallel)
api-designer (SDL/OpenAPI changes) ‖ database-engineer (schema/migrations) — one Task batch after the "Proceed?" question. Results to the user for agreement; then codegen. A BLOCKED from either agent (e.g. database-engineer cannot migrate without a decision) is surfaced at once with a partial report; the slice does not continue on a guessed contract or schema.
Phase 4: Implementation
Backend (go-engineer/php-engineer/node-engineer + graphql-engineer) → once the contract is ready, in parallel frontend (angular-engineer/vue-engineer, css-engineer) and game (threejs-engineer/web-game-engineer) — the frontend may start on mocks from the contract. Engineers write the tests; e2e — test-engineer. Product code and tests are written by these Tier-3 engineers, never by this skill (coordination-rules § Subagents).
Phase 5: Verification
Per the review mode of Phase 1: appsec-engineer (mandatory for sensitive work, else per mode) ‖ accessibility-specialist (UI) ‖ performance-engineer (budget risk) — one Task batch. Findings go to the engineers of Phase 4 for fixes (with consent); open findings are listed in Phase 7. The code review itself runs after the commit, as the hand-off below.
Phase 6: Commit and PR
As /dev-story Phase 6 (git-workflow, step "Implement"):
- Status. On every story card stamped in Phase 2 set
Status: Review(as/dev-storyPhase 6 step 1 —/story-donerequires it), and the session state toNext: /code-review --diffthrough<hooks>session-state.sh; the Phase 2 "start" answer covers this edit, and the card rides the commit below. - Commit gate, one
AskUserQuestion: commit and push (Recommended) · commit only · not now. Before asking, record it —<hooks>session-state.sh set Gate "/team-feature Phase 6: commit?"— and clear it after the answer (<hooks>session-state.sh set Gate "—"), so an open gate survives the next turn (rule 7). - Stage by name. Read
git status --shortfirst and deal with every unplanned??entry (delete a spike, add a file deliberately if it belongs to the slice, otherwise leave it and name it); the story cards of Phase 2 are among the slice's files. Nevergit add -A. git commit -m "feat(<scope>): <feature title>"(feat(S-NNN): …for one story,feat(F-NNN): …for several), thengit push -u origin feat/<scope>-<slug>. Never commit on the default branch.- Find out what starts CI (
grep -l "pull_request" .github/workflows/*.yml, each workflow'son:). When onlypull_requeststarts the checks, open the PR as a draft in this step (gh pr create --draft --fill) so they run during review;/story-donemarks it ready and merges it. Otherwise say which run to expect and its id. - Wait for CI only for a run that exists: one background
gh run watch <run-id> --exit-status; no polling, noAskUserQuestionas a pause.
Phase 7: Summary
Table criteria ↔ tests ↔ results; open findings; the branch, the commit and the PR (or why there is none); propose the code review, then /story-done per story.
Verdict: COMPLETE | PARTIAL | BLOCKED (stage 1: feature-spec <verdict>) | BLOCKED (stage …). Next step — one AskUserQuestion: /web-studio:code-review --diff (copy mode /code-review --diff) — then /story-done per story (Recommended on COMPLETE) · commit first (when Phase 6 was declined) · /story-done per story (when the review already passed) · /release-checklist · fix the open findings first. Run the next skill only on that answer.
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
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.

