Skip to content

feat(skills/product-epic): add standalone epic-scoping skill#19

Merged
sourcehawk merged 2 commits into
mainfrom
feat/product-epic-skill
May 31, 2026
Merged

feat(skills/product-epic): add standalone epic-scoping skill#19
sourcehawk merged 2 commits into
mainfrom
feat/product-epic-skill

Conversation

@sourcehawk
Copy link
Copy Markdown
Owner

Description

Adds product-epic, a standalone skill that takes whatever a product manager has — a rough idea, a few bullets, an early draft — and drives it to an epic an engineer can design against. It sits upstream of planning-a-feature; nothing in the feature-development flow routes into it, and it routes into nothing except writing-github-issues to land the result.

The skill exists because "is this epic ready?" is the wrong question to answer by re-reading your own draft — you hold the whole idea in your head, so it always reads clear. This skill makes readiness something a fresh reader proves, not something the author asserts.

Changes

  • New skill skills/product-epic/SKILL.md and its templates/product-epic.md epic template.
  • Readiness is proven by isolated fresh-reader subagents — an engineer (the primary gate: can they design against this without guessing?) and a stakeholder (do they get the problem and the why?) — given only the epic text.
  • Two hard rules: nothing unconfirmed is written as fact (it becomes an owned ## Open confirmations item), and the scope-vs-usability cut is the PM's call to make and the skill's job to record, not invent.
  • Documented in the README skill table only, matching the existing convention for standalone skills (e.g. drafting-a-release); the project-CLAUDE.md flow diagram is intentionally left unchanged.
  • Plugin version bumped 0.5.0 → 0.6.0 (additive feature).

Testing

Dogfooded end-to-end this session against a real, messy product epic. The skill caught a direct contradiction between the value proposition and the implementation notes, surfaced unsubstantiated SLA/RTO numbers being passed off as fact, and — after a refinement pass — a second isolated fresh-engineer reader confirmed zero surprise blockers remained, with every unknown either resolved or filed as an owned confirmation. The fresh-reader isolation (subagents given only the epic text) behaved as designed: they found gaps the author's own read had glossed.

sourcehawk and others added 2 commits May 31, 2026 22:28
Adds a standalone skill that takes a rough product epic and drives it to
readiness for engineering: it proves readiness with isolated fresh-reader
subagents (engineer + stakeholder) rather than self-assessment, refuses to
write unconfirmed assumptions as fact, and lands the result via
writing-github-issues. Sits upstream of planning-a-feature; nothing in the
feature flow routes into it, so it is documented in the README table only,
matching the existing convention for standalone skills.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
New skill (product-epic) is an additive feature; minor bump.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Copilot AI review requested due to automatic review settings May 31, 2026 20:29
@sourcehawk sourcehawk merged commit 8f7e3c8 into main May 31, 2026
1 check failed
@sourcehawk sourcehawk deleted the feat/product-epic-skill branch May 31, 2026 20:30
Copy link
Copy Markdown
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants