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/paceaitian/paceflow/auditnpx skills add paceaitian/paceflow --skill auditgit clone --depth 1 https://github.com/paceaitian/paceflowWhat 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 | $0.00044 | $0.02512 |
| Opus 5 | $0.00022 | $0.01256 |
| Sonnet 5 | $0.00009 | $0.00502 |
| Haiku 4.5 | $0.00004 | $0.00251 |
Grade A, and why
audit 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 — 164 lines — stays where its author put it; the contents beside it link to each section on GitHub.
PACEflow 全面审查
内部资料：该流程只用于 PaceFlow 仓库自身审计，不随 marketplace 插件发布，也不是用户项目工作流 skill。
触发场景
- 用户说"完整分析"、"全面审查"、"全面检查"
- 用户调用
/audit - 版本发布前的质量门控
审查原则
本 skill 的核心价值是独立发现问题，不是照 guidebook/action-plan/README 打勾。
方法论内核见可发布通用版
review-methodology.md（独立发现 / 证据优先级 / 报告全部再验证 / 三件武器 / 严重度纪律 / 误报防御七条 / 记录基线）。本内部 skill 不重复方法论本体，只在其上叠加 PaceFlow 专用靶子——下面的证据优先级具体路径、审查范围 Glob、五维度并行划分、v7 审计基线。两处共用同一套方法论，避免漂移。
证据优先级：
- 当前代码与配置：
plugin/hooks/**、plugin/agents/**、plugin/agent-references/**、plugin/skills/**、plugin/.claude-plugin/**、.claude-plugin/** - 当前测试与 fixture：
tests/**、tests/agent-tests/** - 真实运行证据：
plugin/hooks/pace-hooks.log、Claude Code session JSONL、production smoke 产物 - 用户面/内部文档：
README.md、REFERENCE.md、CLAUDE.md、docs/**
文档只能用于发现候选矛盾或设计意图，不能单独作为 bug 证据。任何 C/H 级问题都必须从代码路径、配置注册、测试缺口或真实日志中独立证明。
输出范围：
- Phase 1 报告所有可疑发现，不按“看起来低优先级”提前丢弃；严重度只做标注。
- Phase 2 再做证据验证、去重、降级和误报剔除。
- 最终报告必须包含：确认发现、部分正确/有意设计、误报分析、验证矩阵、证据来源、剩余风险。
- 每个发现都写清：文件:行号、触发路径或证据、影响、建议修复；没有代码/测试/日志证据时写成候选或文档问题。
长 diff / stall 防御：
- 大文件或大区间审计时，先用
git log --stat <range> -- <file>定位相关 commit，再用git show <sha> -- <file>逐个查看。 - 不要一次性对
pace-utils.js、pre-tool-use.js、长文档或整个大区间跑无界git diff。 - 长文件读取优先用
rg定位函数/标题，再sed -n读取小范围。
当前 v7 审计基线：
- marketplace
source指向./plugin；发布面是 4 个用户 skill +artifact-writeragent + hooks/agent-references/migrate；internal/skills/audit/、docs、tests、tickets 不随 marketplace 发布 - v7
changes/**详情模型；v5 活跃流程只允许迁移/桥接，不继续兼容 - artifact root 可为 local/vault/custom，真实 git worktree 沿用宿主项目
.pace/artifact-root artifact-writer是唯一 artifact 写入者；主 session 不得直写 C/V/R 标记（含<!-- REVIEWED -->与 frontmatterreviewed-date）- 项目级
artifact-writer.lock串行化 shared artifact 写入；Bash 不得修改该锁 approve-and-start与close-chg是主路径合并操作；验证证据由主 session 运行并读取- review gate 与 V 阶段同构：R 阶段标志
<!-- REVIEWED -->/reviewed-date与 VERIFIED /verified-date平行（action=review↔action=verify），close-chg折叠 VERIFIED 时同样折叠 REVIEWED；agent-lifecycle-guard.js对close-chg与action=review强制review-confirmed/review-source/review-findings（缺则 hard-deny），stop.js对“已 verified 但未 reviewed”给软门提醒；review gate 只记录审计步骤发生+记录，不裁决质量 SubagentStop报告标题问题是观察/恢复提示，不是 artifact 功能阻断
What ships with it
2 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.
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 · 164 lines · 44 tokens per session scan A 5eb7a3e48ae2
audit is a skill published in the GitHub repository paceaitian/paceflow (38 stars, last pushed 15d ago), licensed MIT. It adds 44 tokens to every session and 2,512 once invoked, about $0.0002 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.
Other skills, from other repositories
concord-relay
Keep this Grok session reachable by other Concord agents while idle.
elasticsearch-esql
Generate and execute ES|QL (Elasticsearch Query Language) queries from natural language and visualize results with Vega-Lite charts. Translates user intent into valid ES|QL queries, executes them against Elasticsearch, and can render professional data visualizations (bar, line, scatter, heatmaps, small multiples…
eddo-todo
Todo and task management via Eddo MCP server. Use for tracking work items, remembering tasks, managing project todos and time tracking. Supports GTD workflow with contexts and actionability tags (next actions, projects, waiting, someday). Use this skill whenever you need to remember something, track progress, or…
eddo-work
Structured work mode for implementing complex tasks with phases. Use when user says "what next?" followed by "let's work on this" or "start working". Provides REFINE, IMPLEMENT, REVIEW, COMPLETE, and GITHUBPR phases with gates and reflection checklists. Works with eddo-todo for task tracking.
searxng-search
Web search and content extraction via SearXNG. No API keys, no rate limits. Available in containerized agents via the eddo-chat Docker network.
bcc-plan-spar
BCC align+lock+review PLAN.md for one slice (no product code). Slash: /bcc-plan-spar · chat: bcc:plan-spar · "lock PLAN" · spar the plan. Args: rounds=N (auto-review cap), review=self|subagent|cli|auto|off. Grill until clear enough (no default Q&A quota). Hand off to bcc-clean-cut after human APPROVE.