What local data sources does this skill access?

It analyzes JSON thread files in ~/.local/share/amp/threads/ and history/project JSONL files located in the ~/.claude/ directory.

Is my data shared externally during analysis?

No, all analysis is performed locally on your machine using your stored thread history; no data is sent to external servers for the mining process.

Does this skill require DuckDB?

Yes, it leverages DuckDB for efficient querying of large JSON datasets, enabling fast extraction of message patterns and usage statistics.

Can I use this to find my Amp API key?

Yes, it identifies local storage paths for secrets, settings, and environment variables like AMP_API_KEY and AMP_URL.

How does it help with LLM cost management?

The skill maps your local thread data against current Claude and Gemini pricing models to provide detailed cost profiles and optimization strategies.

Amp API Awareness

Name: Amp API Awareness
Author: plurigrid

byplurigrid

•

分析与监控

Analyzes local thread storage and Claude history via DuckDB to uncover undocumented Amp API patterns and tool usage.

关于

Amp API Awareness is a specialized diagnostic skill designed to provide transparency into the underlying Amp ecosystem by mining local data. It utilizes DuckDB to execute sophisticated queries against thread JSON files and Claude history logs, allowing users to extract hidden API schemas, monitor tool invocation frequencies, and detect active MCP servers. This skill is essential for developers who need to audit agent behavior, reverse-engineer undocumented endpoints, or perform detailed cost-benefit analyses of LLM usage across thousands of local sessions.

主要功能

Detection and mapping of MCP (Model Context Protocol) server activity
Thread schema discovery for structured data extraction and TypeScript interfaces
Automated extraction of tool usage patterns from local thread storage
Detailed DuckDB-powered analysis of JSON and JSONL history files
Comprehensive LLM cost estimation and model distribution mapping
2 GitHub stars

使用场景

Reverse-engineering local storage schemas for custom tool development
Optimizing API costs by analyzing model usage across thousands of messages
Auditing agent tool invocations to identify performance bottlenecks

关于

主要功能

Detection and mapping of MCP (Model Context Protocol) server activity
Thread schema discovery for structured data extraction and TypeScript interfaces
Automated extraction of tool usage patterns from local thread storage
Detailed DuckDB-powered analysis of JSON and JSONL history files
Comprehensive LLM cost estimation and model distribution mapping
2 GitHub stars

使用场景

Reverse-engineering local storage schemas for custom tool development
Optimizing API costs by analyzing model usage across thousands of messages
Auditing agent tool invocations to identify performance bottlenecks