Batch-review meaning index suggestions and auto-accept high-confidence changes.
npx skills add https://github.com/bozaah/meaning-fs --skill meaning-review
Please help me install this Agent Skill. Skill: meaning-review Source: https://github.com/bozaah/meaning-fs/tree/main/.claude/skills/meaning-review Command: npx skills add https://github.com/bozaah/meaning-fs --skill meaning-review
This Skill addresses the time-consuming task of reviewing automated index suggestions and approving changes at scale, ensuring high accuracy while reducing manual effort.
Run /meaning-review to batch-process index suggestions. Use /meaning-review --interactive to review low-confidence items manually.
High-intent search queries and answers about installing and using this skill.
Batch-reviewing codebase index suggestions classifies automated changes into auto-accept, recently added, and manual review groups, auto-accepting high-confidence updates to reduce manual effort while flagging low-confidence items for human review.
Metadata inference for codebase indexing automatically proposes intent updates, tag additions, and relationship mappings across files. Suggestions are generated with confidence scores, and changes meeting a high threshold are applied automatically to the index.
Auto-accepting high-confidence index changes requires a threshold-based flow where suggestions scoring above 0.8 are applied automatically, while lower-scoring items are categorized for manual review to ensure index accuracy and safety.
Interactively reviewing low-confidence index suggestions is supported by running the process with an interactive flag, letting you manually inspect and decide on items below the auto-accept threshold instead of leaving them unprocessed.
Validating index health after applying automated codebase updates happens automatically post-batch, where the system logs all applied changes and runs a validation check to ensure the updated index remains consistent and accurate.
Handling AI-driven codebase indexing at scale without losing accuracy relies on threshold-based batch processing, auto-accepting high-confidence metadata inferences while routing uncertain suggestions to a manual review queue for human oversight.
yuki-aigcInfer env, team, and owner tags for cloud resources using heuristic and topology rules.
bozaahSync the semantic index with filesystem changes by detecting and updating modified files.
bozaahQuery the .meaning/ index with natural language to locate files by intent and tags.
MINT-IAScore and rewrite coaching texts in ARB files with grep/wc metrics.
