Secret scan on tool args: a practical cheatsheet for stopping agent key leaks
blaizedsouza · x · 2026-09-14
- A practical pattern from a developer: agents happily paste secrets into tool arguments, so scan tool args before the call leaves.
- The cheatsheet: detect keys, tokens and private URLs; block the call on a hit; redact args in logs; alert the owning tenant; teach the planner to ask for a safe id instead; test with planted fake secrets.
- Core principle: a tool arg is an egress path. Both model output and user paste can leak—scan both.
More from coding & agent
- AI Filmmaker Spent $4.5k and 847 Takes to Make a 22-Minute Short with Seedance — DavidmComfort · 2026-09-14
- Evomap launches EvoX beta: one swarm agent that codes, researches, and builds slides — CodeByPoonam · 2026-09-14
- Dev Proposes AI-Safety Contractor Board With Cross-Model Verified Receipts — EricBuess · 2026-09-14
- LangChain ships Managed Deep Agents, evolving from LangSmith Deployment into a full agent runtime — hwchase17 · 2026-09-14
- New Podcast with Cloudflare Engineer: AI Coding Agents, Local Models and Agent Harnesses in Practice — Arindam_1729 · 2026-09-14
- Dev on AI coding: explore with ambiguity, but know exactly what to build before building — _Stocko_ · 2026-09-14