skill-self-bootstrap

Retrofit existing OpenClaw skills with self-bootstrap memory capability.

This skill upgrades a target skill so it can:

• read references/LESSONS.md before execution,
• preserve reusable Hard Constraints / Common Failure Cases / Best Practices,
• and continuously improve from prior memory instead of repeating known mistakes.

In short: this is a meta skill for skill infrastructure, not for business task execution.

---

What this skill does

Given one or more target skills, skill-self-bootstrap will typically:

1. Normalize/add a Skill Memory and Iteration Protocol section in target SKILL.md
2. Create or refresh references/LESSONS.md
3. Pull prior signal from memory sources (Mem0 + workspace memory files + target skill docs)
4. Distill findings into reusable lesson categories
5. Keep SKILL.md workflow-focused and move corrective memory into LESSONS.md

---

When to use

Use this skill when the user asks for things like:

• “给这个 skill 加自举能力 / 自迭代能力”
• “加错题集 / lessons / best practices”
• “初始化或刷新 LESSONS.md”
• “把历史踩坑经验沉淀到 skill 里”
• “让这个 skill 能记住之前错误并避免重复”

Do not use it for normal domain work (e.g. data query, design, copywriting). It is specifically for upgrading skill memory architecture.

---

Repository structure

skill-self-bootstrap/
├── SKILL.md
└── references/
    └── templates.md

• SKILL.md: the runnable skill instructions
• references/templates.md: reusable protocol/LESSONS template snippets

---

Requirements

From SKILL.md compatibility:

• read
• write
• edit
• exec
• memory_search
• memory_get

These are required because the skill must inspect/modify files and retrieve memory evidence.

---

Install for OpenClaw

Option A — Local workspace clone (recommended)

From your OpenClaw workspace root:

git clone https://github.com/Shire31/skill-self-bootstrap.git skills/skill-self-bootstrap

Or if directory exists, update:

cd skills/skill-self-bootstrap
git pull

Option B — Git submodule (for pinned version control)

git submodule add https://github.com/Shire31/skill-self-bootstrap.git skills/skill-self-bootstrap
git submodule update --init --recursive

---

How agents should invoke this skill

This skill is auto-triggered via SKILL.md frontmatter description, but agents can also invoke it intentionally when users request lesson/bootstrap enhancements.

Typical operator intent examples

• “给 customer-account-research-feishu 加自举能力并初始化 lessons”
• “把 xxx skill 的历史失败经验整理到 references/LESSONS.md”
• “刷新 yyy skill 的 protocol + lessons，按最新记忆修正”

---

Expected output after a successful run

For each target skill, agents should produce:

1. Updated SKILL.md with Skill Memory and Iteration Protocol
2. Updated references/LESSONS.md with reusable entries
3. A short summary including:
   • what changed,
   • where memory was sourced,
   • what was seeded,
   • and any known gaps for next pass

---

Special case: skill-creator

When target is skill-creator, this skill should enforce that:

• newly created skills include self-bootstrap capability by default, unless user opts out;
• new skills should include protocol block + references/LESSONS.md + first-pass lessons initialization.

This turns self-bootstrap into a default factory standard for future skills.

---

Quick verification checklist

After installation, verify:

• skills/skill-self-bootstrap/SKILL.md exists
• skills/skill-self-bootstrap/references/templates.md exists
• agent has access to required tools (memory_search, memory_get, file tools)

Then run a small test prompt:

“帮我给某个已有 skill 补自举 protocol，并生成 first-pass LESSONS.md”

If target skill files are updated with protocol + lessons structure, installation is valid.

---

Versioning and updates

Suggested release flow:

git tag v0.1.0
git push origin main --tags

For consumers:

cd skills/skill-self-bootstrap
git fetch --tags
git checkout v0.1.0

---

License

MIT (see LICENSE).