File-First Agent Orchestration. Readable agents. Inspectable state. Predictable costs.
This repository also configures its own agents. See what claude-pipe tells them →
Agent Claude Code
Classify a batch of items according to a predefined schema. Each worker processes one batch independently, enabling parallel execution across multiple batches.
Agent Claude Code
Write a single final markdown documentation file from structured YAML data. Applies anti-cringe and code-grounding to produce clean, accurate documentation.
Agent Claude Code
Document a single module by reading its source files, extracting public APIs with full signatures, and generating structured documentation data.
Agent Claude Code
Research a single aspect of the topic using web search. Evaluate source quality, extract findings with full attribution.
Agent Claude Code
Generate a comprehensive research report from synthesized findings. Every claim must be traceable to sources.
Agent Claude Code
Scan the target codebase for vulnerabilities in a single category. Use grep patterns to find candidates, read code to confirm in context, eliminate false positives, and produce evidence-grounded findings.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: