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 majiayu000/spellbook --skill capability-distillgit clone --depth 1 https://github.com/majiayu000/spellbookWrote 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/majiayu000/spellbook/capability-distill)<a href="https://agentmods.dev/skills/majiayu000/spellbook/capability-distill"><img src="https://agentmods.dev/badge/skills/majiayu000/spellbook/capability-distill/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/majiayu000/spellbook/capability-distill"><img src="https://agentmods.dev/badge/skills/majiayu000/spellbook/capability-distill.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00136 | $0.02704 |
| Opus 5.5 | $0.00054 | $0.01082 |
| Sonnet 5.5 | $0.00027 | $0.00541 |
| Haiku 4.5 | $0.00014 | $0.00270 |
Grade A, and why
capability-distill 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 2mo 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 — 173 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Capability Distill
把“为什么在两个合理选项中选择其中一个”提取成证据支持的判断规则。不要把操作步骤、通用建议或原始会话内容换个格式包装成 skill。
与现有技能的边界
capability-distill:选择获准证据、还原决策事件、提取判断规则、输出 judgment packet。skill-audit:判断这些规则是否值得成为 skill、归属哪个现有 skill、是否重复以及如何分层。skill-creator:创建或修改具体的SKILL.md、设计 with-skill/baseline eval、迭代和验证触发描述。skill-lifeguard:当产物属于高影响工作流时补可靠性契约和漂移修复闭环。
不要在本 skill 中复制后三者的完整写作、注册或评测流程。目标是提供它们可消费的高信号输入。
Operating Contract
- Direct actions: 使用当前对话和用户明确批准的本地材料;先读元数据再读内容;生成脱敏的 judgment packet;运行本地只读检查。
- Escalate before: 读取未获准的会话历史、memory、home 目录或其他仓库;把任何轨迹内容发送给外部模型;覆盖已有 skill;写入共享仓库;发布或安装产物。
- Evidence-backed pushback: 判断密度低、证据不足或已有 skill 完整覆盖时,给出具体重叠项并建议不蒸馏或只更新原 skill。
- Feedback loop: 把 eval 中的“未生效 / 机械化 / 误触发”连同对应
rule_id回写到下一轮 judgment packet,而不是只改措辞。
0. 建立数据边界
在读取额外材料前记录以下字段:
source_scope:
approved_roots: []
approved_artifact_types: []
external_model_destination:
raw_content_authorized: false
output_path:
当前对话和用户本次明确附带的文件可直接使用。其他路径、历史日志和 memory 不因“可能有帮助”而自动进入范围;缺少批准时先询问并暂停对应读取。
执行以下数据纪律:
- 不假定
~/.claude、~/.codex或任何固定运行时路径存在。使用用户给出的路径、当前工作区和当前运行时可用的搜索工具。 - 先查看文件名、时间、提交主题等元数据,只对候选决策事件读取最小必要片段。
- 不把 token、密钥、cookie、个人身份信息、客户数据、私有源码或完整 prompt/response 写入 packet 或 eval。
- 如需外部强模型,只发送经用户批准的脱敏场景摘要;没有目标模型和数据发送授权就标记阻塞,不假装完成窗口蒸馏。
- provenance 只写可验证的来源标签和日期。模型名、版本或作者未知时留空,不猜测。
1. 判断是否适合蒸馏
先区分对象:
- 可由固定命令或 checklist 完成的是流程,交给
skill-audit/skill-creator,不做判断蒸馏。 - 需要根据上下文在多个合理选项间权衡,并且存在切换、停止或上抛信号的,才是候选判断。
- 只包含“谨慎、验证充分、保持简洁”等通用态度时,直接判为低判断密度。
对候选场景记录 judgment_density、evidence_strength 和 existing_coverage(high / medium / low)。只保留判断密度高、证据至少中等且现有覆盖不完整的场景。
2. 还原决策事件
从获准证据中提取事件摘要,而不是复制原文。每个事件至少包含:
decision_event:
evidence_ref:
context:
viable_options: []
chosen_option:
observed_signal:
outcome:
counterfactual:
redactions_applied: []
evidence_ref 使用本地、非敏感的定位信息,例如仓库相对路径和 commit SHA;不要把原始聊天文本塞进该字段。没有 outcome 的事件只能作为假设,不能升级为高置信规则。
3. 写作前重叠审计
在写任何 judgment_rule 前,先在 source_scope 获准的 roots 或已附 inventory 内搜索现有 rules、skills、registry 和相近 owner;先查索引和元数据,再读最可能相关的最小片段。没有获准的 owner 根目录或清单时,先请求范围并暂停规则起草,不能把“未提供”当作“无覆盖”。
What ships with it
5 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.
- 2mo ago First seen · 173 lines · 136 tokens per session scan A 661da634f632
capability-distill is a skill published in the GitHub repository majiayu000/spellbook (288 stars, last pushed yesterday), licensed MIT. It adds 136 tokens to every session and 2,704 once invoked, about $0.0005 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
cao-session-management
Find available CAO (CLI Agent Orchestrator) agent profiles, and talk to the conductor and worker terminals of a running CAO session, including unblocking a stuck worker. Use when choosing a profile to launch, messaging or unblocking workers, or diagnosing a stuck session. For preflight, launching, session commands…
workflow
A workflow skill for embedded-device projects that plans and runs building, flashing, debugging, observing, and diagnosis tasks.
pneuma-session
Instructions for renaming an active Pneuma session and replacing its default preview with a useful title and summary. A Pneuma session is one work area inside a larger project.
fast-resume
Search local coding-agent session history and identify safe resume commands. Use when the user asks to find, inspect, continue, or recover previous work from Claude Code, Codex, Pi, or another agent indexed by fast-resume.
lov-organize-storage
A storage-organization tool for surveying a whole drive or archive and planning a project-based folder layout. It can rename items within the same drive after confirmation, while keeping a mapping, log, and rollback script.
lov-finder-action
A generator for custom right-click actions in macOS Finder, the file manager on Apple computers.