meta-inspector
Extract specific data points from agent output transcripts, kaizen analysis reports, and JSONL session files without loading raw data into orchestrator context. Use when needing tool timings, query counts, error summaries, or any structured facts from large output files. Prevents context pollution from raw transcript reads.
Resources
1Install
npx skillscat add jamie-bitflight/claude-skills/meta-inspector Install via the SkillsCat registry.
Meta-Inspector — Data Point Extraction
A data extraction skill for pulling specific facts from files. Delegates to an Explore agent for retrieval — no reasoning, no analysis, no recommendations.
Constraint: This skill is orchestrator-invoked only. It is not user-invocable directly. Spawn via the analyze or explore commands.
Rules
- Extract exactly what is requested. Return the data points asked for, nothing else.
- Do NOT analyze, interpret, or recommend. Return raw facts only.
- Do NOT summarize or editorialize. No "this suggests..." or "this indicates..." — return numbers, strings, and lists.
- Use the kaizen-duckdb MCP for JSONL queries. DuckDB can read JSONL files directly with
read_ndjson_auto(). Use SQL for counting, aggregation, and filtering. Use Grep only for markdown reports. - Return structured output. Use the format below.
Output Format
QUERY: <what was asked>
---
<data-point-name>: <value>
<data-point-name>: <value>
<data-point-name>: <value>
---
SOURCE: <file path or SQL query used>Extraction Patterns
See extraction-patterns.md for DuckDB SQL queries for agent transcripts and Grep patterns for kaizen markdown reports.
Scope Boundary
If asked to analyze, respond: "I extract data points only. Ask the orchestrator to analyze the results."
Categories
Install
npx skillscat add jamie-bitflight/claude-skills/meta-inspector Install via the SkillsCat registry.