Self-Improving to ExpertPack

Convert Self-Improving Agent learnings into a structured ExpertPack. Migrates the .learnings/ directory (LEARNINGS.md, ERRORS.md, FEATURE_REQUESTS.md) and an...

Security Scan

The skill's code and instructions match its stated purpose (converting a .learnings/ directory into an ExpertPack); it requires only python3, reads local workspace files, and writes output to a user-specified directory—no network calls or unrelated credentials requested.

Purpose & Capability

Name/description promise (migrate .learnings into ExpertPack) aligns with the included script and SKILL.md usage. Required binary (python3) is appropriate and no unrelated credentials or host access are requested.

Instruction Scope

SKILL.md instructs running the bundled convert.py against a workspace and writing an output ExpertPack. The script reads expected files (LEARNINGS.md, ERRORS.md, FEATURE_REQUESTS.md, promoted files like CLAUDE.md) within the provided workspace and writes structured files to the provided output path—behavior consistent with the stated task. The SKILL.md does not instruct reading unrelated system files or sending data externally.

Install Mechanism

No install spec (instruction-only), but the skill includes a runnable Python script. This is low-risk and expected for this kind of utility, but because code is bundled it will be executed on the host filesystem when invoked—users should inspect the script before running.

Credentials

The skill requests no environment variables, no credentials, and no config paths. The script does attempt to detect and redact common token patterns in files it reads (e.g., sk-, ghp_, xoxb-), which is appropriate for its purpose.

Persistence & Privilege

always is false and the skill does not request persistent system-level privileges or modify other skills. It writes output only to the user-specified output directory; this level of filesystem access is proportional for a conversion tool.

Assessment

This skill appears coherent and performs local file conversion. Before running it: (1) review scripts/convert.py yourself to confirm it matches your expectations (it is bundled in the skill), (2) run it against a copy or in a sandbox to avoid accidental overwrite, (3) back up your workspace first, and (4) inspect the produced ExpertPack before publishing—its built-in secret-stripping covers common patterns but may not catch every secret, so do not rely on it as the only protection.

Self-Improving Agent → ExpertPack

Converts a Self-Improving Agent skill's .learnings/ directory (3.8K ClawHub installs) into a properly structured ExpertPack.

Supported sources:

• LEARNINGS.md — corrections, knowledge gaps, best practices, simplify-and-harden patterns
• ERRORS.md — command failures, exceptions, integration issues
• FEATURE_REQUESTS.md — user-requested capabilities and implementation notes
• Promoted content — entries already promoted to CLAUDE.md, AGENTS.md, SOUL.md, TOOLS.md (detected and cross-referenced)

Usage

cd /root/.openclaw/workspace/ExpertPack/skills/self-improving-to-expertpack
python3 scripts/convert.py \
  --workspace /path/to/your/workspace \
  --output ~/expertpacks/my-learnings-pack \
  [--name "My Agent's Learnings"] \
  [--type auto|person|agent|process]

Override .learnings/ location with --learnings /path/to/.learnings.

What It Produces

A complete ExpertPack conforming to schema 2.3:

• manifest.yaml (with context tiers, EK stub)
• overview.md summarizing conversion (entry counts, categories, priority breakdown)
• Structured directories mapped from learning types:
  • mind/ — best practices, conventions, behavioral patterns, promoted rules
  • facts/ — knowledge gaps filled, project-specific facts
  • operational/ — error resolutions, tool gotchas, integration fixes
  • summaries/ — pattern analyses, recurring issue summaries
  • relationships/ — cross-references between related entries
• _index.md files, lead summaries, glossary.md (if terms/tags found)
• relations.yaml (from See Also links and shared tags)
• Clean deduplication preferring promoted > resolved > pending entries

Secrets are automatically stripped (sk-, ghp_, tokens, passwords). Warnings emitted for any found.

Post-Conversion Steps

1. cd ~/expertpacks/my-learnings-pack
2. Run the ExpertPack chunker: python3 /path/to/expertpack/scripts/chunk.py --pack . --output ./.chunks
3. Measure EK ratio: python3 /path/to/expertpack/scripts/eval-ek.py .
4. Review overview.md and manifest.yaml
5. Commit to git and publish to ClawHub

Learn more: https://expertpack.ai • ClawHub expertpack skill

See also: Self-Improving Agent skill on ClawHub.