spec-execution

A workflow that turns one written specification—such as a requirements document or GitHub issue—into one reviewable pull request.

In plain words
What is it for?
Use it to implement a defined change, run the repository’s required tests and checks, and arrange one final code review.
Why use it?
It keeps implementation within the agreed scope and attaches proof that the required checks passed. TDD, or test-driven development, means using tests as part of the implementation loop.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/assisjp/specflow/spec-execution
Any agent
npx skills add assisjp/specflow --skill spec-execution
Clone the repo
git clone --depth 1 https://github.com/assisjp/specflow

Made for: Claude Code, Codex.

Per session 103 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,398 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00103 $0.03398
Opus 5 $0.00051 $0.01699
Sonnet 5 $0.00021 $0.00680
Haiku 4.5 $0.00010 $0.00340

Measured 2d ago against content hash 9b1aaaf1df64, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

spec-execution scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/engineering/spec-execution/SKILL.md · 128 lines

How it starts

The opening of the file, as written. The whole thing — 128 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Spec Execution

Turn one specification into one reviewable PR.

This skill is a sequence, not a rulebook. The prohibitions (--no-verify, disabling lint, the attempt limit) live in the repo's verification block (AGENTS.md or CLAUDE.md, per ADR 0003) and hold always. Here is only the order of things.

It does not reimplement TDD or review — it calls tdd and code-review.

Human review happens once, on the final diff.

1. Contract

Read the verification block from the repo's agent-context file — AGENTS.md, or CLAUDE.md if that is where the harness loads it and the block lives there. Extract the canonical commands, including the Run command for the evidence gate.

The file must be in the repository — tracked by git, in this working tree. A user-level or global CLAUDE.md is not it, however faithfully the harness loads it: it describes its author's habits across every project, not this repo's contract, and the commands it names (npm test, npm run lint) routinely do not exist here. Borrowing them is the improvisation this step exists to prevent, wearing the costume of a contract. git ls-files decides what counts, not find.

If no block exists: stop. Suggest running repo-hardening first. Do not improvise commands, install tooling, or continue.

2. Read before writing

Read the whole source — a spec, a PRD, or a ticket (in no-tracker mode you are run once per ticket from docs/tickets/<slug>/, and that ticket is the spec for this PR) — every file it names, the files those import, and CONTEXT.md if it exists, before editing a single line. Note the source's path; step 7 hands it to code-review.

First, check the source for a return marker — a returned label on the issue (gh issue view <n> --json labels), or a Returned: line in the local ticket/spec file. Read it per issue, never through a gh issue list --label returned filter: GitHub's label index lags writes by seconds (to-tickets hits the same lag after bulk creation), so a filtered read can miss a marker that was just set and wave a defective spec straight through. If the source is marked, it was already sent back as a spec defect and not yet healed: do not implement — go straight to the second-failure rule below. The marker is what makes that rule a gate instead of relying on someone remembering (see the rule for why). The way out is to heal the source and republish it — to-spec for a spec, to-tickets for a ticket — because republishing is what clears the marker (ADR 0008). If the source has not actually been rewritten yet, that healing starts with a fresh grill.

Read the full file on GitHub · 128 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 128 lines · 103 tokens per session scan A 9b1aaaf1df64

Subscribe to this mod's changes

spec-execution is a skill published in the GitHub repository assisjp/specflow (17 stars, last pushed 22d ago), licensed MIT. It adds 103 tokens to every session and 3,398 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.

obra/superpowers · 21 tokens

next-cache-components-adoption

Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…

vercel/next.js · 95 tokens

babysit-pr

Babysit a GitHub pull request after creation by continuously polling review comments, CI checks/workflow runs, and mergeability state until the PR is merged/closed or user help is required. Diagnose failures, retry likely flaky failures up to 3 times, auto-fix/push branch-related issues when appropriate, and keep…

openai/codex · 114 tokens

imagegen

Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output…

openai/codex · 113 tokens

cpu-profile-analysis

Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…

microsoft/vscode · 71 tokens

next-cache-components-optimizer

Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…

vercel/next.js · 170 tokens