Self Improving To Expertpack
PassAudited by ClawScan on May 1, 2026.
Overview
This appears to be a purpose-aligned local converter, but it packages persistent local learning files for reuse, so users should review the generated ExpertPack before sharing or publishing it.
Use this only on workspaces you intend to export. After conversion, inspect the generated ExpertPack for private information, secrets that the simple redaction patterns may not catch, and instructions you do not want future agents to follow. If PyYAML is needed, install it from a trusted, preferably pinned source.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Private or stale agent learnings could be preserved and reused in future contexts, especially if the generated pack is committed or published.
The skill intentionally converts accumulated agent learnings and promoted instruction files into a persistent retrieval pack, which may later influence agents or expose private workspace knowledge if shared.
Migrates the .learnings/ directory ... and any promoted content from workspace files into ExpertPack's portable format with multi-layer retrieval, context tiers
Review the generated files for private data, obsolete instructions, and unsafe guidance before using, committing, or publishing the ExpertPack.
The tool may fail until PyYAML is installed, and an unpinned manual package install relies on the user's package source and environment.
The script has a runtime PyYAML dependency, but the supplied requirements only declare python3 and there is no pinned install spec.
import yaml ... print("Error: pyyaml required. Install with: pip install pyyaml", file=sys.stderr)Install PyYAML from a trusted package index, preferably with a pinned version in a controlled environment.