Does Meta-Inspector analyze the extracted data?

No. Meta-Inspector is designed strictly for factual data extraction to prevent context pollution. It returns raw facts, and any interpretation or analysis is performed by the orchestrator.

Can I use this skill to query SQL databases?

While it uses DuckDB, its primary purpose is querying structured log files like JSONL and session transcripts rather than general-purpose database management.

How do I trigger the Meta-Inspector skill?

It is an orchestrator-invoked skill. It is not meant to be called directly by the user but is spawned via the 'analyze' or 'explore' commands during a workflow.

How does it handle very large log files?

It uses DuckDB to query JSONL files directly (using read_ndjson_auto) and Grep for markdown, allowing it to filter and aggregate data without loading the entire file into the LLM's memory.

Meta-Inspector Log & Data Extractor

Name: Meta-Inspector Log & Data Extractor
Author: Jamie-BitFlight

byJamie-BitFlight

•

分析と監視

Extracts precise data points and performance metrics from large agent transcripts and JSONL logs without polluting the model's context.

Meta-Inspector is a specialized utility designed to retrieve structured facts, tool timings, and error summaries from massive session files and kaizen analysis reports. By leveraging DuckDB for JSONL processing and Grep for markdown, it allows Claude to access critical data points—such as query counts or specific values—without loading raw, high-token-count logs into the orchestrator's primary context. This ensures efficient processing and prevents context window bloat during complex agentic workflows by providing only the specific facts required for the task at hand.

主な機能

01Strictly factual output format to prevent AI hallucination or editorializing

02High-performance SQL querying via DuckDB for session data aggregation

0326 GitHub stars

04Automated extraction of tool execution timings and query counts

05Context-efficient data retrieval from large JSONL and markdown files

06Precise Grep-based pattern matching for structured kaizen reports

ユースケース

01Summarizing error counts across multiple JSONL log files using SQL queries

02Auditing agent performance by extracting execution times from long session transcripts

03Isolating specific data points from kaizen reports for cross-session comparisons

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jamie-bitflight/claude_skills meta-inspector

For use in Claude.ai and ChatGPT

Download Skill