We got local models to triage the OpenClaw repo for FREE!*

Summary

OpenClaw receives hundreds of issues and pull requests daily, creating a triage problem for maintainers who need near-real-time filtering without consuming a paid closed-model quota. The system uses Gemma and Qwen local models inside a Pi-based agent harness: each item is normalized, enriched with GitHub context, classified into finite topic labels, and required to return structured output. For pull requests, the agent can inspect the repository through reposhell, a read-only command-restricted shell, while localpager, SQLite, a worker queue, and deterministic notification policy handle orchestration and Discord delivery. Tests described for gemma-4-26b-a4b and qwen3.6-35b-a3b report hundreds of tokens per second and good zero-shot classification accuracy without fine-tuning, while an LLM audit loop monitors errors; the authors present the pattern as agentic classification for high-throughput triage.

Takeaways

1. 1

   Reposhell limits the model to read-only repository commands such as ls, find, rg, grep, sed, cat, and git inspection; unsupported commands such as curl are rejected.

2. 2

   The architecture keeps classification agentic but makes notification deterministic: the model labels each item, while localpager applies the user's topic policy before sending Discord alerts.

3. 3

   An audit loop used roughly 40,000 GPT-5.5 tokens per two-hour check, costing about 2–3 cents per run or approximately $9 per month at 12 runs per day.