AI Delivery Spec
Turns a loose feature request into a gated, traceable requirement brief or PRD
Test report
- Verdict
- Tested · Works
- Score
- Tested
- Aug 10, 2026
- Environment
- Claude Code 2.x (agent harness)
Wrote the same 'add CSV export to the orders list' brief twice and ran the skill's own Python gate on both: the no-skill version came back BLOCKED, the skill-guided one came back REVIEW_COMPLETE_WITH_GAPS with zero blockers and three open unknowns that each carry an owner, a blocking stage and a fallback. The content gap was bigger than the format gap — the skill version caught CSV formula injection, row-count parity against the list counter, a separate export permission and an audit record, none of which the plain brief mentioned, and it converted three ownerless 'open questions' into tracked items. The gate is honest about its own limits: every run prints NOT_PROVEN for domain correctness, runtime behaviour and real acceptance. The catch is language — SKILL.md and the README are roughly ninety percent Chinese, and parts of the CLI help are too, so English-only teams read the manual through a translator.
Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works
- Installs cleanly 5/5
- Triggers reliably 4/5
- Output vs. baseline 8/10
- Docs & honesty 4/5
What AI Delivery Spec does
A requirements-management kernel for Claude Code: it routes a request to the right stage (frame, explore, intake, clarify, specify, review, baseline, change, acceptance) and delivers the smallest complete artifact for that stage, with stable REQ/DEC/UNK ids and owned open questions. Bundled Python gates then check the artifact statically and refuse to call structural conformance business correctness. Triggers on writing, changing, reviewing or accepting requirements, PRDs and prototypes; it explicitly excludes coding, CI/CD and deployment.
How to install AI Delivery Spec
git clone https://github.com/franklinxkk/ai-delivery-spec.git
mkdir -p ~/.claude/skills
cp -r ai-delivery-spec ~/.claude/skills/ai-delivery-spec
Skills live in ~/.claude/skills/ (global) or .claude/skills/
(per-project). Restart Claude Code after installing.
Commands — how to trigger AI Delivery Spec
-
/ai-delivery-specTurns a loose feature request into a gated, traceable requirement brief or PRD
It also activates on plain-language prompts like these:
-
Write a clarify-stage requirement brief for a CSV export button -
Draft a PRD with REQ/DEC/UNK ids for this new feature request -
Move this feature from intake to specify stage with open questions
Frequently asked questions
- Is the AI Delivery Spec skill free?
- Yes. The skill itself is free from franklinxkk/ai-delivery-spec. SkillProof publishes the install command and an independent test verdict at no cost.
- Does AI Delivery Spec work with Claude Code?
- We tested it with Claude Code 2.x (agent harness) on Aug 10, 2026. Verdict: Tested · Works. Wrote the same 'add CSV export to the orders list' brief twice and ran the skill's own Python gate on both: the no-skill version came back BLOCKED, the skill-guided one came back REVIEW_COMPLETE_WITH_GAPS with zero blockers and three open unknowns that each carry an owner, a blocking stage and a fallback. The content gap was bigger than the format gap — the skill version caught CSV formula injection, row-count parity against the list counter, a separate export permission and an audit record, none of which the plain brief mentioned, and it converted three ownerless 'open questions' into tracked items. The gate is honest about its own limits: every run prints NOT_PROVEN for domain correctness, runtime behaviour and real acceptance. The catch is language — SKILL.md and the README are roughly ninety percent Chinese, and parts of the CLI help are too, so English-only teams read the manual through a translator.
- What is the AI Delivery Spec SkillProof Score?
- 8.4/10 — installs cleanly 5/5, triggers reliably 4/5, output vs. baseline 8/10, docs & honesty 4/5.
- How do I install AI Delivery Spec?
- Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
- Can I use AI Delivery Spec with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
- The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.