@wanng-ide

auto-skill-hunter scans unresolved problems and session history to discover, rank, and install relevant skills from clawhub. execution is bounded by install caps and includes a dry-run safety mode.

hle-benchmark-evolver ingests benchmark results into a curriculum generation pipeline for capability-evolver, producing reward signals and prioritized question queues to guide hle score optimization.

json-modifier applies RFC 6902 patches to JSON files with atomic writes and pre-flight validation. covers config updates, package.json mutations, and array modifications without full file rewrites.

json-repair-kit normalizes malformed json through node.js evaluation, handling trailing commas, single quotes, unquoted keys, and comments. creates backups by default and validates output before writing.

arxiv-agentic-verifier generates discriminative test cases to expose code logic flaws by reasoning about problem constraints rather than random sampling, supporting python and javascript.

markdown-linter validates markdown files for broken local links, missing references, and basic syntax issues. targets workspace documentation integrity but lacks procedural depth and trigger specificity.

arxiv-skill-learning periodically fetches papers from specified categories, extracts skill code via external extractor, runs tests, and commits results to workspace. no details on failure handling or extractor reliability.

extracts reusable skill code from arxiv papers via automated pipeline, but lacks concrete implementation details, error handling, and clear trigger conditions.

Builds a reusable, scored memory mesh with safety gating and 12-hour auto-refresh for cross-session memory consolidation and quality control in OpenClaw.

Patrol latest arXiv papers and auto-generate Node.js learned skills through hunter to extractor pipeline.