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.
git clone --depth 1 https://github.com/ronronner02/codepilot-agentWrote 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/commands/ronronner02/codepilot-agent/spec-write-skill)<a href="https://agentmods.dev/commands/ronronner02/codepilot-agent/spec-write-skill"><img src="https://agentmods.dev/badge/commands/ronronner02/codepilot-agent/spec-write-skill/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/ronronner02/codepilot-agent/spec-write-skill"><img src="https://agentmods.dev/badge/commands/ronronner02/codepilot-agent/spec-write-skill.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00013 | $0.02087 |
| Opus 5.5 | $0.00005 | $0.00835 |
| Sonnet 5.5 | $0.00003 | $0.00417 |
| Haiku 4.5 | $0.00001 | $0.00209 |
Grade A, and why
spec-write-skill 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 1mo 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 — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Command support root: .claude/spec-first/workflows/spec-write-skill. Treat it as the loaded skill directory whenever this inlined workflow refers to SKILL_DIR or the directory containing SKILL.md.
Spec Write Skill
把可复用目标转成 portable、source-first 的 Skill patch，或在零写入模式下报告 package readiness；交付正确分支的结果、匹配证据与 residual risks，不把 source bytes、fixture pass 或模型自述当成语义改善。
Scenario Capability
Follows docs/contracts/workflows/scenario-capability-matrix.md (default).
Overrides: none
Workflow Contract Summary
- Input / output: 用户目标、一个 target repo/Skill root、现有 package、项目规则、相邻 Skill、已接受 findings 和必要只读参考，产生 near-neighbor route、
validate-onlyreport、preview 后的单 repo source patch 或 source-resolution blocker；每个结果带验证状态与 residual risks。 base_operation=create|revise只区分新建 package 与处理现有 package；effect=apply|validate-only决定副作用，只有effect=apply才允许修改已确认的 canonical source，effect=validate-only即使面对现有或外部 package 也保持零写入。modifier=migrate|audit-remediation|none只补充输入分析，不形成新 workflow/effect。layer_result是 runtime 输出合同：near-neighbor-route|refuse-generated-runtime-patch|portable-core-only|portable-core-with-behavior-contract|portable-readiness-report|trust-preflight-blocked|blocked-source-owner|spec-first-project-profile。新增值必须同步更新 source、consumer 和 tests，不能只写入 maintainer fixture。
Branch Contract
先根据用户请求和已确认事实选择一个 disposition；只读取会改变该 disposition 判断的 reference，并在下列 done signal 达成后停止。
| Disposition | Entry signal and result | Required action / evidence | Done signal and failure behavior |
|---|---|---|---|
| Near-neighbor | 非 authoring/readiness，或只请求 audit-only quality review、纯安装、runtime mirror。已接受 finding 的 remediation 不属于本分支。输出 base_operation=null、effect=not-entered、modifier=none；结果为 near-neighbor-route 或 refuse-generated-runtime-patch。 |
只给 owning route / next action；安装交给 skill-installer，mirror 交给 runtime-maintenance。 |
路由后停止；不得 inventory、validator、preview 或 mutation。 |
| Owner blocked | create/revise 的 owner 不唯一、跨 repo、repo-external、generated-only 或 containment 未确认。保留 `base_operation=create | revise与effect=apply，结果为 blocked-source-owner`。 |
读 Authoring Method，给 candidate-only preview、空 would-change/command list 和唯一下一步。 |
| Validate-only | 用户明确检查现有/外部 package；现有 package 使用 base_operation=revise + effect=validate-only，结果为 portable-readiness-report 或 trust-preflight-blocked。 |
no-follow inventory、bundled validator 与 Delivery Gates。 | 报告后停止；不得执行目标 scripts、validator、hooks、binaries 或 lifecycle，不得跟随 symlink、读 secret-like 内容、复制、安装或写入。 |
| Tier A apply | 已确认 owner 的 behavior-preserving revise，具体条件由 workbench 定义。 | 读 Authoring Workbench，确认当前授权覆盖 exact write set、preview binding 与最窄结构验证。 | receipt/验证后 close out；承重行为变化转 full apply。 |
| Full apply | 已确认 owner 的 create/revise apply，结果为 portable-core-only、portable-core-with-behavior-contract 或 spec-first-project-profile。 |
依次读 Authoring Method、Authoring Workbench 与 Shape-Aware Evaluation Design；写 core 前完成紧凑 Design Record 与最小 pre-patch eval plan。Capability Map、显式 shape/module decision 或 topology 只在它们改变 owner、consumer、resource/runtime carrier、架构或风险时展开。 | preview、授权、风险匹配验证和 source update 后 close out；缺 baseline/eval plan 时停止，缺 semantic/comparative evidence 时降级对应 claim。 |
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.
- 1mo ago First seen · 61 lines · 13 tokens per session scan A 5369718702c9
spec-write-skill is a command published in the GitHub repository ronronner02/codepilot-agent (0 stars, last pushed 1mo ago), licensed MIT. It adds 13 tokens to every session and 2,087 once invoked, about $0.0001 per session on Opus 5.5. 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 commands, from other repositories
infogenius
Research-First Image Generation - Facts → Visual Prompt → High-Quality Image.
close-track
Close a completed track — run quality gate, update conductor state, commit, and optionally handle the git branch.
audit-plugin
Audit plugin skills, commands, and agents for structure, size, and naming issues.
release
Prepare a release (step 1 of 2) — version bump and PR. Cutting the tag is a separate operator-only step.
new-agent
Scaffold a new LangGraph agent with best practices.
just-ship-vps
VPS verwalten — Status pruefen, Projekte verbinden, neuen VPS einrichten.