CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

wf-plan

This skill should only be used when the user uses the word workflow and asks to create a commit-ready, deep project plan from a shaped packet (brief, breadboard, risks, spikes) before development starts, with handoff/pickup boundaries to avoid context rot.

Quality

78%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./.agents/skills/03-plan/wf-plan/SKILL.md
SKILL.md
Quality
Evals
Security

wf-plan

Purpose

Turn a shaped packet (wf-shape dossier) into a commit-ready plan without re-litigating the shape.

Never implement production code. Only research and write the plan.

Inputs

Dossier folder path containing:

  • brief.md
  • breadboard-pack.md
  • risk-register.md
  • spike-investigation.md (if present)

Outputs

Inside the same dossier:

  • plan.md

Steps

  1. Pickup (recommended if new thread)
  • Invoke pickup if repo/branch state is not fresh.
  1. Ingest shaped packet (no idea refinement loop)
  • Read brief → breadboard → risk register → spikes.
  • If perimeter or top risks are missing: route back to wf-shape.
  1. Create plan skeleton
  • Create plan.md with:
    • scope (in/out)
    • key flows + key logic
    • acceptance criteria + verification plan
    • sequencing/phases + stop points
    • rollout/rollback + observability
    • dependencies + risks
  1. Local research (always)
  • Find similar patterns in repo.
  • Pull institutional learnings (docs/solutions, docs/LEARNINGS.md).
  • Record concrete file paths.
  1. External research (conditional)
  • Always external for: security/auth, payments, privacy, external APIs, migrations.
  • Otherwise only if local context is thin.
  1. Spec hardening (gap pass)
  • Edge cases, failure modes, concurrency, performance, data integrity, security threats.
  • Update acceptance criteria + verification.
  1. Plan review passes
  • Simplicity, risk, ops/release, data integrity, security/privacy, UX/product.
  • Final mandatory pass: invoke oracle on the whole plan and integrate findings (or mark out-of-bounds).
  1. Commit gate
  • GO only if:
    • AC measurable
    • verification per AC
    • sequencing explicit
    • rollout/rollback explicit
    • no P1 unknowns remain
  1. Handoff to build (recommended boundary)
  • Invoke handoff and include:
    • plan path + summary of phases
    • how to verify
    • rollout notes
    • biggest remaining risks
  • Recommend build in a fresh thread:
    • /new then pickup then run wf-develop or wf-ralph

Verification

  • plan includes acceptance criteria + verification + rollout/rollback
  • oracle pass integrated

Go/No-Go

  • GO if the plan can be built without re-discovering the shape.
  • NO-GO if it depends on “we’ll figure it out during implementation”.
Repository

marchatton/agent-skills

Path

.agents/skills/03-plan/wf-plan/SKILL.md

Last updated
First committed

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.