agentleFS
theyoungastronautstheyoungastronauts/polaris/skills/full-feature/SKILL.md

full-feature

Skill4 starsChanged 3 months agoMITView on GitHub

End-to-end feature workflow — plan, scaffold, execute, verify, and merge, phase by phase.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

14 rules. About 0 are common, 9 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈9

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If the feature has a user-facing interface:

    line 15
  2. Very specific:

    Run /prd to formalize requirements into a structured PRD

    line 16
  3. Very specific:

    Run /ux-spec to design the experience through 6 UX passes (mental model, IA, affordances, cognitive load, state design, flow integrity)

    line 17
  4. Very specific:

    Use the planner agent

    line 21
  5. Distinctive:

    If this is a new project with sub-repos to create:

    line 28
  6. Very specific:

    Use /scaffold from the root planning folder

    line 29
  7. Distinctive:

    Start a Claude session in the sub-project

    line 40
  8. Distinctive:

    Start a NEW Claude session in the same directory

    line 45
  9. Distinctive:

    If this phase produced API changes, generate the integration summary for frontend phases

    line 53
  10. Distinctive:

    Single feature: check out a branch, execute, review, PR, merge.

    line 64
  11. Very specific:

    Multiple independent features in parallel: use git worktrees (see the /worktrees skill).

    line 66
  12. Distinctive:

    Don't skip the planning step — it saves time overall

    line 69
  13. Distinctive:

    Keep phases small enough to review in one sitting

    line 70
  14. Distinctive:

    Update the plan as you learn — it's a living document

    line 72

≈ “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

Workflow: Full Feature

Overview

End-to-end workflow for implementing a feature from plan to merged code.

Steps

1. Product Definition (optional, for UI-heavy features)

If the feature has a user-facing interface:

  • Run /prd to formalize requirements into a structured PRD
  • Run /ux-spec to design the experience through 6 UX passes (mental model, IA, affordances, cognitive load, state design, flow integrity)
  • Skip this step for backend-only or technical features

2. Plan

  • Use the planner agent
  • Skills loaded: plan-and-scope, phase-breakdown
  • The planner will consume PRD and UX spec artifacts if they exist
  • Output: plan.md with phased tasks
  • Review and iterate on the plan before proceeding

2b. Scaffold (new projects only)

If this is a new project with sub-repos to create:

  • Use /scaffold from the root planning folder
  • Creates sub-project directories, git inits, installs profiles
  • Skip this step for features in existing repos

3. Execute — MVP Build

For initial builds where you're taking a plan from zero to working application, work directly on main. No branches or worktrees needed — there's nothing to protect yet.

Repeat for each phase:

Execute:

  • Start a Claude session in the sub-project
  • Tell it: "Execute phase N of the plan"
  • The executor agent reads plan.md and implements all tasks — it does not commit; changes stay uncommitted until verification passes

Review:

  • Start a NEW Claude session in the same directory
  • Tell it: "Review phase N against the plan"
  • The reviewer agent checks the work and produces a verification report
  • Fix any FAIL items, re-verify if needed

Commit and move on:

  • On PASS, the review session commits the phase (implementation + verification report) to main — the session that owns the work commits (in the hands-off flows below, that's the lead)
  • Between phases, the commit history provides a natural review point
  • If this phase produced API changes, generate the integration summary for frontend phases
  • Continue to the next phase

Hands-off alternatives:

  • /autopilot — for pre-planned work. Executes all plan.md phases: implement → test → verify → commit. Stops on FAIL — run /autopilot N to resume.
  • /orchestrator — for broader work. Accepts inline tasks or plan.md, supports parallel execution, auto-phasing, and per-task model overrides. Use when tasks need decomposition or research.

4. Execute — Ongoing Feature Work

Once you have a working application and are building new features on top of it:

Single feature: check out a branch, execute, review, PR, merge. Simple.

Multiple independent features in parallel: use git worktrees (see the /worktrees skill). Each feature gets its own worktree so you can switch between them without stashing or losing context. This is where worktrees earn their keep.

Tips

  • Don't skip the planning step — it saves time overall
  • Keep phases small enough to review in one sitting
  • The verification step catches things you'd miss in self-review
  • Update the plan as you learn — it's a living document

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.