v0.1.7

What's Changed

• build(deps): Bump vite by @dependabot[bot] in #36
• build(deps): Bump lodash-es from 4.17.23 to 4.18.1 by @dependabot[bot] in #38
• build(deps-dev): Bump flatted from 3.4.1 to 3.4.2 by @dependabot[bot] in #37

Full Changelog: v0.1.6...v0.1.7

v0.1.6

What's Changed

• Add edge type descriptions in Japanese to Knowledge Graph UI by @chigichan24 in #15
• Replace logo and unify brand colors to amber by @chigichan24 in #16
• Update screenshot by @chigichan24 in #17
• Fix tokenize() RangeError on large session corpora by @chigichan24 in #25 Thank you for your report @daniswhoiam in #18

Full Changelog: v0.1.5...v0.1.6

v0.1.5

Changes

• Improve skill generation quality by leveraging /insights facets data (#13)
  • Merge narrow clusters based on shared goal categories
  • Add successRate / helpfulness signals to reusability scoring
  • Add a Session Insights section to the synthesis prompt
  • Automatically run /insights during analyze-sessions
• Use underlying_goal as the basis for topic labels
• Show facets brief_summary on the session list and rename the column to "Summary"
• Prevent synthesis-time JSONL session files via --no-session-persistence
• Filter out claude -p-originated sessions with isNonInteractiveSession()
• Strip synthesis output preambles via stripSynthesisPreamble()
• Reset synthesis state when switching topics in the Knowledge Graph
• Add publishConfig for scoped package public access

New CLI flags

Flag	Description
--facets-dir <path>	Custom facets directory (default: ~/.claude/usage-data/facets)
--skip-facets	Skip /insights refresh and facets integration

Full Changelog: v0.1.4...v0.1.5

v0.1.4

Changes

• Bump synthesis timeout from 120s to 300s (applied to both CLI and UI)
• Add npm badges and a CLI usage section to the README

Full Changelog: v0.1.3...v0.1.4

v0.1.3

crune v0.1.3

Initial release. crune is a CLI tool that generates reusable skill definitions from Claude Code session logs.

Features

• Generate skills via npx @chigichan24/crune (no clone required)
• Full pipeline: session parsing → knowledge graph construction → skill candidate detection → LLM synthesis
• Output follows the Claude Code skill specification (<name>/SKILL.md)

Usage

npx @chigichan24/crune --dry-run                     # preview candidates
npx @chigichan24/crune --skip-synthesis              # heuristics only
npx @chigichan24/crune --count 3 --model haiku       # with LLM synthesis
npx @chigichan24/crune --output-dir ~/.claude/skills # install directly

npm

https://www.npmjs.com/package/@chigichan24/crune