epic
Skill0 starsChanged 5 months agoMITView on GitHub
Break a spec or feature into prioritized, code-grounded epics — ordered by risk and dependency, each with acceptance criteria and a sequencing diagram.
5 other projects use a version of this.
Adopted by
Most recently taken up by
- jamesm-dev/tateru.devtheir own version, first seen Nov 1, 2026
- nexus-a1/claude-skillstheir own version, first seen Oct 17, 2026
- ulises-jeremias/agent-toolkittheir own version, first seen Aug 31, 2026
- Code-Shock/claude-skillstheir own version, first seen Apr 5, 2026
- skilluse/skillusetheir own version, first seen Mar 3, 2026
Other versions5 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 5
- skilluse/skilluse★ 55same name
- ulises-jeremias/agent-toolkit★ 20same name
- Code-Shock/claude-skills★ 5same name
- jamesm-dev/tateru.dev★ 1same name
- nexus-a1/claude-skills★ 1same 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.
The file
We split this skill into the rules it gives an agent
9 rules. 0 are common, 7 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 5 edited versions we read keep these.
7
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
- Distinctive: line 14
Code-grounded only: Only reference code, patterns, or risks you have actually read.
- Distinctive: line 22
Read CLAUDE.md if it exists.
- Distinctive: line 24
Ask for the input if the user hasn't provided one:
- Distinctive: line 27
Read the relevant areas of the codebase that the feature will touch.
- Distinctive: line 36
Cohesive: all stories share a common deliverable goal
- Distinctive: line 41
If an epic violates one of these properties, reshape or split it — and say why.
- Distinctive: line 115
After producing the complete epic breakdown, save it to the project:
- Very specific: line 117
Create docs/epics/ if it doesn't exist.
- Very specific: line 119
Write the full output to: docs/epics/YYYY-MM-DD-{slug}.md using today's actual date.
“Common” is measured: kept by at least 6 in 10 of the 5 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.
Files
/epic — Epic Breakdown
You are running the urdesigner /epic skill. Your job is to break a spec or feature description into a set of prioritized epics — cohesive, independently deliverable units of work grounded in the actual codebase.
Ground rules
- Planning only. You will not write, modify, or suggest code changes.
- Code-grounded only. Only reference code, patterns, or risks you have actually read.
- Honest uncertainty. Mark "NEEDS SPIKE" where the evidence is too thin to define a scope — don't make up scope.
- 3–7 epics. Fewer is better. Resist the urge to over-decompose.
Step 1 — Gather context
- Read
CLAUDE.mdif it exists. - Read
STRATEGY.mdif it exists. - Ask for the input if the user hasn't provided one:
- A completed
/specoutput, or - A feature description (in which case, ask clarifying questions before proceeding if the scope is ambiguous).
- A completed
- Read the relevant areas of the codebase that the feature will touch. Identify integration points, existing complexity, and where the risk lives.
Step 2 — Identify and shape epics
What makes a good epic
An epic is:
- Cohesive: all stories share a common deliverable goal
- Independently shippable: it can go to production without requiring another epic to ship first
- Right-sized: 1–4 weeks of work for a senior engineer (not a sprint, not a quarter)
- Testable: clear acceptance criteria at the epic level
If an epic violates one of these properties, reshape or split it — and say why.
Epic prioritization criteria (apply in order)
- Risk first: epics that de-risk unknown technical or UX bets ship first. Spikes are their own epic.
- Dependencies second: epics that others depend on ship earlier.
- User value third: epics that unlock the most visible user value ship before polish.
- Complexity last: purely additive or cosmetic epics ship when the risky work is done.
Step 3 — Output
Produce 3–7 epics using this format:
Epics: [Feature Name]
Input: [spec title or feature description] Files reviewed: [list key files read]
Epic 1: [Name]
Priority: 1 (Highest) Rationale: Why this ships first — reference specific risk or dependency observed in the codebase.
Scope What this epic includes. Be specific — reference actual code areas, data models, or components.
Out of Scope What this epic explicitly does NOT include.
Acceptance Criteria
- Criterion (testable, observable)
- ...
Dependencies
- Depends on: none / [Epic N]
- Blocking: [Epic N] / none
Rough size: XS / S / M / L / XL (Signal only — not story points. XS ≈ 1–3 days, S ≈ 1 week, M ≈ 2 weeks, L ≈ 3–4 weeks, XL = needs splitting)
[Repeat for each epic]
Risks and Unknowns
List any unknowns that could change epic scope. For each:
- What is unknown
- Why it matters to scope
- What a spike would answer
- Which epic it blocks
Mark items "NEEDS SPIKE" that must be resolved before the blocked epic is ready to schedule.
Sequencing
Epic 1 ──→ Epic 2 ──→ Epic 4
└──→ Epic 3 ──→ Epic 5Explain any non-obvious ordering decisions below the diagram.
Save artifact
After producing the complete epic breakdown, save it to the project:
- Create
docs/epics/if it doesn't exist. - Derive a kebab-case slug from the feature name (e.g. "Checkout Flow" →
checkout-flow). - Write the full output to:
docs/epics/YYYY-MM-DD-{slug}.mdusing today's actual date. - Tell the user the exact path where it was saved.
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.

