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 skills add arch-team/devpace --skill pace-learngit clone --depth 1 https://github.com/arch-team/devpaceWrote 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/arch-team/devpace/pace-learn)<a href="https://agentmods.dev/skills/arch-team/devpace/pace-learn"><img src="https://agentmods.dev/badge/skills/arch-team/devpace/pace-learn/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/skills/arch-team/devpace/pace-learn"><img src="https://agentmods.dev/badge/skills/arch-team/devpace/pace-learn.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 | $0.00071 | $0.00948 |
| Opus 5.5 | $0.00028 | $0.00379 |
| Sonnet 5.5 | $0.00014 | $0.00190 |
| Haiku 4.5 | $0.00007 | $0.00095 |
Grade A, and why
pace-learn 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.
What it actually says
/pace-learn — 经验积累与知识管理
devpace 的学习引擎。双模式运行：
- 自动模式（主体）：在关键事件后静默提取经验 pattern 到
.devpace/metrics/insights.md - 手动模式：用户通过子命令管理知识库
子命令路由
重要：根据触发方式（自动/手动）和子命令，仅读取路由表中对应的 procedures 文件，不加载其他 procedures。
| 子命令 | 说明 | 详见 |
|---|---|---|
| （无参数/自动触发） | 事件驱动的自动学习 | 本文件"自动模式流程" + skills/pace-learn/learn-procedures.md |
note <描述> |
手动沉淀探索中的有价值经验 | skills/pace-learn/learn-procedures-query.md §1 |
list [--type TYPE] [--tag TAG] [--confidence MIN] |
按条件筛选 pattern | skills/pace-learn/learn-procedures-query.md §2 |
stats |
知识库统计概览 | skills/pace-learn/learn-procedures-query.md §3 |
export [--path FILE] |
导出可复用经验 | skills/pace-learn/learn-procedures-query.md §4 |
prune [--dry-run] [--decay-only] |
清理低置信度/过期条目 | skills/pace-learn/learn-procedures-query.md §5 |
自动触发场景
| 触发事件 | 事件类型 | 学习重点 | 提取目标 |
|---|---|---|---|
| CR merged | merged |
成功模式 | 可复用的检查项、高效的工作路径 |
| Gate fail | gate1_fail / gate2_fail |
失败教训 | 检查项阈值调整、Claude 盲区识别 |
| 人类打回 | rejected |
理解差距 | 意图理解偏差模式、审查标准校准 |
| 挣扎信号 | struggle |
环境缺陷识别 | Skill/procedures/Schema 改进建议 |
触发源由 hooks/post-cr-update.mjs 检测并输出 devpace:learn-trigger 提醒。
自动模式流程
重要：自动模式按以下步骤执行，仅读取 skills/pace-learn/learn-procedures.md 中对应 Step，不加载 skills/pace-learn/learn-procedures-query.md。
Step 1：前置检查
检查 .devpace/ 是否存在——不存在则静默退出。存在则按触发源确定提取方向（merged→成功 pattern | Gate fail→防御 pattern | rejected→改进 pattern）。
Step 2：提取 Pattern
读取 skills/pace-learn/learn-procedures.md Step 2 执行。
Step 3：对比与积累
读取 skills/pace-learn/learn-procedures.md Step 3 执行。
Step 4：嵌入式反馈
读取 skills/pace-learn/learn-procedures.md Step 4 执行。
输出
- 自动模式：嵌入式 1 行通知（有新 pattern 时），或静默（无有价值发现时）
- 手动模式：按子命令输出，详见
skills/pace-learn/learn-procedures-query.md
约束
- 不修改 CR 文件或其他状态文件（仅写入 insights.md）
- 不中断当前工作流
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.
- 1mo ago First seen · 68 lines · 71 tokens per session scan A 22020dd6fcec
pace-learn is a skill published in the GitHub repository arch-team/devpace (73 stars, last pushed 6mo ago), licensed MIT. It adds 71 tokens to every session and 948 once invoked, about $0.0003 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 2026-08-30.
Other skills, from other repositories
query
Use when asking questions about the project wiki, querying accumulated knowledge, searching project decisions or lessons. Triggers include "query the wiki", "what does the wiki say about", "search project knowledge", and natural language questions directed at the field-guide wiki.
retro
Use when completed work needs to be recorded as a lore artifact covering what happened, what surprised, what broke, and what drifted. Observation only, not interpretation. Triggers include "let's do a retro", "/retro", "review what happened", "capture what happened", and "write up the session".
new
Start a new session — save summary to memory, then mark as reset so next message gets a fresh greeting. OpenClaw-compatible. Triggers on /new, /reset, /agent:new, "nueva sesión", "new session", "reset".
alive-morning
Morning briefing -- read all walnut states, surface priorities, inbox count, stale walnuts, people nudges.
fresh-start
Continue in a new chat without losing anything — writes a compact handoff to paste into a fresh conversation. Use whenever the user wants to move to a new chat or conversation, hits usage limits, says replies got slow or Claude forgets earlier details, or the conversation has grown long.
token-check
Estimate how many tokens a pasted text, an attached file or the conversation so far takes, what it means for the context window and usage limits, and how to cut it. Use when the user asks how big something is in tokens, why they keep hitting limits, or whether a file is too large to send.