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/macintog/codex-spine/project-continuitynpx skills add macintog/codex-spine --skill project-continuitygit clone --depth 1 https://github.com/macintog/codex-spineWhat 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 | $0.00084 | $0.02883 |
| Opus 5 | $0.00042 | $0.01442 |
| Sonnet 5 | $0.00017 | $0.00577 |
| Haiku 4.5 | $0.00008 | $0.00288 |
Grade A, and why
project-continuity 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.
How it starts
The opening of the file, as written. The whole thing — 153 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Project Continuity
Keep a long-lived project aligned on purpose while narrow implementation work changes. The goal is a compact, discoverable continuity contract, not more documentation.
Concrete bootstrap tools, memory retrieval, Git mechanics, indexing, and local proof commands belong to the installed environment or a tooling guide the repository explicitly claims. Do not infer a repo-local codex/TOOLING.md or a Git policy from this skill.
Workflow
- Bind the latest explicit user-selected task subject before reading a packet as actionable: objective and scope, canonical repository or system, exact worktree/ref/HEAD or runtime identity, and current authority/evidence. A goal, queue, checkpoint, prompt, packet, index hit, or worker output is evidence only until every binding field matches.
- Infer the mode:
inspectis read-only;adoptestablishes a packet;repairseparates mixed roles;resumereconciles the handoff with the bound subject and current evidence;handoffrefreshes current state and archives displaced detail. - Establish the authority scope: one repository, a multi-repository workspace, or one product or program spanning repositories. Start unfamiliar ownership at
undetermined; then establishrepo-native-only,local-overlay, orrepo-localfrom evidence. Treatin-tree-adoptionandrepo-native onlyonly as deprecated read aliases; never write them back. - Inspect the applicable
AGENTS.mdandAGENTS.override.mdinstruction chain for the current working directory, native documentation and source-of-truth files, existing continuity and ownership declarations, repository and publication posture, and conflicting, duplicated, stale, generated, or sensitive material. Stay read-only until the current task and repository policy authorize writes. - Reconcile each fact into exactly one role: durable product authority, repo-local working rule, advisory volatile handoff state, durable decision or deep reference, or historical evidence. Treat every checkpoint and state anchor as a claim about prior state, not proof or task selection.
- Apply the smallest idempotent change. Preserve truthful native guidance, merge rather than replace existing agent rules, avoid duplicating current state, and make no changes when the existing packet is healthy.
- Verify that a fresh agent can discover the packet without knowing this skill exists, pointers resolve, current-state claims match repository or runtime evidence, template instructions and secrets are absent, and only the intended startup files load routinely. Run
scripts/audit-continuity.py --root <path> --jsonwhen this skill's bundled auditor is available. Astaleorfailresult blocks resume and handoff; it is never authority to write or to follow another file's next step. - Return the mode, scope, posture, inspected and changed files, authority conflicts resolved, verification performed, and unresolved decisions or missing authority. End terminally: residuals are non-directive findings, not a generated successor task.
What ships with it
10 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.
- agents/openai.yaml 314 B
- assets/AGENTS.fragment.md 2.2 KB
- assets/ARCHIVE_NOTE.template.md 811 B
- assets/CHECKPOINT.stub.template.md 296 B
- assets/CHECKPOINT.template.md 2.2 KB
- assets/PROJECT_CHECKPOINT.model.json 228 B
- assets/PROJECT_CONTINUITY.template.md 2.1 KB
- references/adoption-procedure.md 4.6 KB
- references/self-hosting-signposts.md 1.8 KB
- scripts/audit-continuity.py 28 KB runs code
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.
- 2d ago First seen · 153 lines · 84 tokens per session scan A 885bee059215
project-continuity is a skill published in the GitHub repository macintog/codex-spine (9 stars, last pushed 2d ago), licensed MIT. It adds 84 tokens to every session and 2,883 once invoked, about $0.0004 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
bensz-rmd-rules
规范 AI 开发 R Markdown 分析脚本的行为准则。当用户要求"写 Rmd 分析"、"开发 R 脚本"、"做数据分析"时触发。核心原则:遵循主业与副业分离架构(.R 保留完整数据,.Rmd 应用业务阈值),优先使用用户已有 R 包资源;图表默认按 Nature 级别可读性与出版质量生成;专家级解读兼顾弱背景读者,提供四层框架、指标导读与不常用指标首次解释协议;路径验证确保跨平台兼容性。前提:luckyBase 为硬依赖。.
auto-test-project
当用户明确要求"测试项目"、"运行 auto-test-project"或"进行项目级测试"时使用。对完整项目进行多轮 A 轮批判性测试 + B 轮质量检查,系统化发现、记录、修复问题。⚠️ 不适用:用户只是想优化功能(应直接修改)、只是询问项目问题(应直接回答)、没有明确"测试"意图。.
auto-test-code
当用户明确要求"测试代码"、"运行代码审查"或"进行代码自检"时使用。通过多轮 A 轮批判性代码审查 + B 轮代码质量原则检查,系统化发现、记录、修复程序代码中的问题,并将计划/过程/结果统一沉淀到目标代码根目录的 .bensz-api/task-{yyyymmdd-hhmm}-{简短描述}/auto-test-code/{yyyy-mm-dd-hh-mm}/output/tests/ 隔离工作区。⚠️ 不适用:用户只是想优化功能(应直接修改)、只是询问代码问题(应直接回答)、没有明确"测试代码"意图。.
auto-test-skill
当用户明确要求"测试技能"、"运行 auto-test"或"进行批判性测试"时使用。通过多轮 A 轮批判性测试 + B 轮质量原则检查,系统化发现、记录、修复问题,并沉淀可追溯的 .bensz-api/task-{yyyymmdd-hhmm}-{简短描述}/auto-test-skill/output/plans/ 与 .bensz-api/task-{yyyymmdd-hhmm}-{简短描述}/auto-test-skill/output/tests/ 文档。⚠️ 不适用:用户只是想优化功能(应直接修改)、只是询问技能问题(应直接回答)、没有明确"测试"意图。.
frontend-specialist
前端开发专家。用于前端应用、页面、组件、现有 UI 增量改进、已有页面视觉优化、UI/UX 改造、视觉重设计、响应式布局、表单和输入框视觉一致性、设计系统落地、React/Vue/Next.js/Nuxt、Tailwind、shadcn/ui、Radix、MUI、AntD、仪表盘、SaaS、Landing Page、移动端适配、前端性能优化、浏览器截图检查和去除 AI 味 UI。触发后应把“做好看”转化为可执行设计约束,优先复用项目现有组件与视觉语言,小步实现并尽量通过浏览器或 Playwright 做视觉反馈闭环;当用户要改进当前页面或已有界面时,默认先诊断再做局部优化,而不是从 0 重做。.
git-pr-review
当用户明确要求“review 某个 GitHub PR”“评估某个 pull request 是否值得 merge”“帮我判断这个 PR 怎么处理”时使用。基于用户提供的 GitHub 仓库地址、PR 编号/链接和补充说明,进行只读、证据驱动的 PR 审查:理解 PR 解决的问题、评估方案优劣与局限、默认优先使用内置“好 PR”标准并在必要时联网补充、识别恶意或高风险改动,并输出是否建议 merge 的 Markdown 决策报告。⚠️ 不适用:用户要你直接修改 PR 代码、直接 merge PR、或在本地执行 PR 分支中的不可信代码。.