Skip to content

feat(ce-prototyping): add PM prototyping skill#1072

Open
nstrimple wants to merge 4 commits into
EveryInc:mainfrom
nstrimple:feat/ce-prototyping-skill
Open

feat(ce-prototyping): add PM prototyping skill#1072
nstrimple wants to merge 4 commits into
EveryInc:mainfrom
nstrimple:feat/ce-prototyping-skill

Conversation

@nstrimple

@nstrimple nstrimple commented Jul 7, 2026

Copy link
Copy Markdown

What & why

Adds ce-prototyping: a product-manager-facing skill that turns an idea into a working prototype and a highly complete, validated PRD for engineering. It attacks the product→engineering "game of telephone" — imperfect, incomplete communication of intent — so teams build the right thing the first time with less back-and-forth.

How it works

  • Opening setup — adapts to the PM's technical level, then has them choose a point on a fidelity spectrum (UI-only demo → production-seed) and the environment (standalone artifact vs in-repo), reconciling a production-seed choice against a missing repo (scaffold or downgrade) rather than failing.
  • Co-evolution loop — build a prototype slice, review it with the PM, refine the requirements, repeat. The prototype is a requirements-discovery tool.
  • Decision log — a running log keeps the prototype and PRD in bidirectional sync, so nothing decided at the demo is lost from the PRD and no requirement contradicts the prototype.
  • Product-principles gate — a shipped baseline (data capture, measurability, success metrics, target user, non-goals, edge/error states, privacy) plus reusable, cross-repo org-specific principles (repo copy + user-global store) must be addressed before a PRD is "complete."
  • Dual PRD output — a canonical ce-unified-plan requirements artifact engineering runs /ce-plan on, plus a readable HTML view rendered from it (so the two can't drift). The prototype is attached as validating evidence.

Changes

  • skills/ce-prototyping/SKILL.md + references/{opening-setup,scoping-dialogue,prototype-build,product-principles,decision-log,prd-output}.md (new, self-contained).
  • docs/skills/ce-prototyping.md (doc page), plus both README inventory tables, the docs/skills/README.md catalog, and the release-metadata skill count (29 → 30).
  • docs/plans/2026-07-07-002-feat-ce-prototyping-skill-plan.md — the ce-brainstorm → ce-plan artifact behind this change.

Validation

  • bun test — 1835 pass, 0 fail
  • bun run release:validate — in sync (30 skills)
  • Behavioral contract (AE1–AE7 in the plan) is designed for skill-creator evals per AGENTS.md "Validating Agent and Skill Changes."
  • Self-review via ce-code-review applied 5 fixes (decision-log phase/location, principles read-order consistency, paraphrase tightening, README count sync).

Note: CI on a first-time fork contribution will show action_required until a maintainer approves the workflow run; local gates pass.

🤖 Generated with Claude Code

https://claude.ai/code/session_01WusDJRhfKHboKJvjF869Cx


CI status

Local gates pass (bun test 1835/1835, bun run release:validate in sync, 30 skills). CI is action_required — awaiting maintainer approval to run workflows on a first-time fork contribution.

nstrimple and others added 2 commits July 7, 2026 19:22
Add ce-prototyping: a product-manager-facing skill that turns an idea into a
working prototype and a highly complete, validated PRD for engineering,
closing the product-to-engineering intent gap.

- Opening setup: adapt to the PM's technical level; choose a fidelity from
  UI-only demo to production-seed and the environment (standalone vs in-repo),
  reconciling a production-seed choice against a missing repo.
- Co-evolution loop: build a prototype slice, review, and refine requirements,
  with a decision log keeping prototype and PRD in bidirectional sync.
- Product-principles gate: a shipped baseline plus reusable, cross-repo
  org-specific principles must be addressed before a PRD is final.
- Dual PRD output: a canonical ce-unified-plan artifact engineering runs
  /ce-plan on, plus a readable view rendered from it.

Registered in both README inventories, the docs/skills catalog and a doc page,
and the release-metadata skill count (30).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Create the decision log in Phase 3 (scoping), where its first entries are
  produced, instead of Phase 4.
- Move the per-run decision log under the .context/ scratch namespace per
  AGENTS.md, leaving the config-like product-principles repo copy in place.
- Tighten the Phase 2 trigger so it stops paraphrasing (and contradicting) the
  reference's principles read-order/interview condition.
- Trim the Phase 5 dual-PRD paraphrase to a load-bearing trigger.
- Sync the README "ships N skills" prose to 30.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

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

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 0d029bf5bd

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

Comment thread skills/ce-prototyping/references/prd-output.md
Comment thread skills/ce-prototyping/references/product-principles.md Outdated
…covery

Codex PR review (EveryInc#1072):
- Resolve the product-principles repo copy (and the decision-log scratch dir)
  at the git root via `git rev-parse --show-toplevel` so a run from a
  subdirectory doesn't read/write a different .compound-engineering file and
  later root/subdir runs miss the active principles.
- Teach ce-plan Phase 0.2 recent-plan discovery (and the Phase 5.1.5 gate) to
  recognize product_contract_source: ce-prototyping, so a prototyping PRD is
  discoverable by a bare /ce-plan, not only via an explicit path.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

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

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 1a7cf459e3

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

Comment thread skills/ce-prototyping/references/prd-output.md
…ance

Codex PR review (EveryInc#1072): after teaching ce-plan to enrich ce-prototyping PRDs,
ce-doc-review's premise-validation gate still recognized only
ce-brainstorm/legacy-requirements provenance, so a plan derived from a
prototype-validated PRD was treated as greenfield and could trip premise-level
adversarial techniques. Add product_contract_source: ce-prototyping to the
validated-upstream signals in SKILL.md and references/subagent-template.md.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
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.

1 participant