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.
npx agentmods add skills/yannlugrin/claude/specifynpx skills add yannlugrin/claude --skill specifygit clone --depth 1 https://github.com/yannlugrin/claudeWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/yannlugrin/claude/specify)<a href="https://agentmods.dev/skills/yannlugrin/claude/specify"><img src="https://agentmods.dev/badge/skills/yannlugrin/claude/specify.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00233 | $0.07927 |
| Opus 5 | $0.00117 | $0.03964 |
| Sonnet 5 | $0.00047 | $0.01585 |
| Haiku 4.5 | $0.00023 | $0.00793 |
Grade A, and why
specify 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 5d 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.
How it starts
The opening of the file, as written. The whole thing — 590 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Specify
Produce and maintain a SPECIFICATIONS.md: a statement of intent and
constraints for a project, written to be implemented by an AI (typically
Claude Code) without babysitting. A specification here is never an
implementation guide — it states what must exist and why, and leaves file
layouts, tool syntax and code to the implementer. Its only reader is the
implementer; documents for anyone else are deliverables of the
implementation, named in the spec but not written there.
Three roles, never blurred:
- The user is the sole arbiter. Every decision and every review finding passes through them. You propose with a recommendation and reasoning; they rule; only then do you apply.
- You (the main session) are the editor and keeper of intent. You hold the history of why things are the way they are, draft and amend the document, triage findings, and maintain the decision log.
- Cold subagents are the reviewers. Fresh context is what makes their reviews worth having; they see only the documents, never this conversation.
The specification is written in English. Converse with the user in whatever language they use.
Workspace
Work inside a git repository (offer git init if there is none):
SPECIFICATIONS.md— the deliverable, at the repository root..claude/spec-work/decisions.md— the decision log (format below)..claude/spec-work/reviews/— archived review reports,NNN-<lens>.mdin chronological order..claude/refs/— reference material the user supplies that is not part of the specification (a contract of a system that will consume the deliverable, an inventory, a document from elsewhere). Committed, and kept out of.claude/docs/: the implementation phase sweeps that directory as its own memory and would eventually fold or delete whatever sits there. Read-only in every phase, this one included — only the user changes what they supplied; a reference that looks wrong is reported, never corrected. Named in the handoff, never in the specification — the spec states requirements, not where to read background..gitignore— carries.claude/worktrees/from setup, before the first review is ever spawned. Worktree isolation materializes the reviewer's checkout inside the repository; a commit made while one exists otherwise swallows it (this has happened). Ignored is not gone: prune the checkouts when the phase ends (git worktree prune, then remove the directory), before the handoff and its squash. Measured on one run: sixteen of them, eight megabytes, still registered against the repository's former path, and still holding copies of the.claude/refs/material the tree had deliberately dropped so the implementer would never receive it..claude/settings.json— written and committed at workspace setup with"autoMemoryEnabled": false(merged in, never overwriting other keys, if the file exists). Auto memory is machine-local and unversioned — an ungoverned second memory outside git and outside review; neither the spec sessions nor the future bootstrap session may accumulate or load it. The implementation workflow keeps it off (its permission-baseline step extends this same file).
What ships with it
19 files 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.
- references/challenger.md 4.4 KB
- references/handoff-assets/approve-step.md 11 KB
- references/handoff-assets/bash_guard.py 101 KB runs code
- references/handoff-assets/code-reviewer.md 5.6 KB
- references/handoff-assets/handover-step.md 7.4 KB
- references/handoff-assets/optimize-memory.md 12 KB
- references/handoff-assets/orient.md 3.6 KB
- references/handoff-assets/resume-step.md 6.0 KB
- references/handoff-assets/state-reviewer.md 8.3 KB
- references/handoff-assets/step-reviewer.md 7.6 KB
- references/handoff-assets/test-reviewer.md 6.7 KB
- references/handoff.md 118 KB
- references/reviewer.md 12 KB
- references/structure.md 10.0 KB
- references/updates.md 14 KB
- references/updates/2026-08-17.md 44 KB
- references/updates/2026-08-19.md 18 KB
- references/updates/2026-08-20.md 13 KB
- references/updates/2026-08-21.md 16 KB
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.
- 5d ago First seen · 590 lines · 233 tokens per session scan A 08ae837ec162
specify is a skill published in the GitHub repository yannlugrin/claude (2 stars, last pushed 10d ago), licensed MIT. It adds 233 tokens to every session and 7,927 once invoked, about $0.0012 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 2026-08-31.
Other skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
local-ai-agents
Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…
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…
next-partial-prefetching-adoption
Turn on Partial Prefetching in a Next.js app and work through the insights it surfaces. Use when the user wants to enable or adopt Partial Prefetching, flip the partialPrefetching flag, opt routes in with export const prefetch = 'partial', audit Link prefetch={true} behavior, preserve existing prefetched UI with…
chronicle
Analyze Copilot session history for standup reports, usage tips, session search, and session reindexing. Use when the user asks for a standup, daily summary, usage tips, workflow recommendations, wants to search or find past sessions by keyword/file/PR, wants to reindex their session store, or asks about deleting…
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…