agentic-gate: open-source layer blocks LLM agents' bad tool calls, 150→0 in benchmark
Nice_Current_8581 · reddit · 2026-10-25
Developer released agentic-gate (TS + Python, MIT), a validation layer sitting between the model and your tools to catch hallucinated arguments like nonexistent order IDs, negative amounts, or vague dates.
How it works
- Validates arguments against Zod (TS) or Pydantic (Python) schemas before execution
- Sends validation errors back to the model so it can fix and retry
- Trips a circuit breaker on repeated failures, avoiding 50-turn retry loops
- Provider-agnostic: adapters for OpenAI, Anthropic, Bedrock, Gemini, plus an async validate() hook
Benchmark
Using a scripted model (measuring the gate, not any specific LLM) across refund, payment transfer, and DB query scenarios: 150 bad calls got through without the gate, 0 with it; model turns dropped from 150 to 12. Scripts in benchmarks/ are reproducible.
The README notes it only checks argument shape and doesn't replace permissions or review of risky actions. Install via npm i agentic-gate / pip install agentic-gate.
More from coding & agent
- Streamrunner ships: AI-powered native macOS reimagining of OBS Studio for $39.99 — jasonkneen · 2026-10-25
- Dev shares crude cross-interface agent setup: Linux box with limited apps plus CUA monitoring — tokenbender · 2026-10-25
- Dev uses agents to monitor group chats and send greetings on his behalf, eyeing voice cloning next — tokenbender · 2026-10-25
- There's no moat in code anymore: what's left is distribution, taste, community — paw_lean · 2026-10-25
- MOSS-Image prompt enhancer packaged as a local Ollama model for offline ComfyUI use — Ducth_IT · 2026-10-25
- Dev gives himself 100 days to build a business on HyperFrames: 20M videos rendered in 30 days — toolstelegraph · 2026-10-25