ATO - Agentic Tool Optimization

[ATO - Agentic Tool Optimization @agentictool]

Receipts agent call — did it use the tools, did it follow your rules — as a column you can query, not a vibe. We documented the whole build, including a regression we shipped ourselves and a claim we later retracted when real data killed it.
Full story 👇
https://t.co/hgJflddW6w

[ATO - Agentic Tool Optimization @agentictool]

Your coding agent has access to your repo. It’s supposed to read the actual files before answering.

The receipt says success. The answer looks right.

But it never opened the file — it answered from training priors. Confidently. And you’d never catch it from the logs.

[ATO - Agentic Tool Optimization @agentictool]

We proved it on ourselves. Same prompt, two models, a repo with no src/auth.ts:
→ one checked the directory and refused to fabricate
→ one wrote 5,454 chars of vulnerability report for a file that doesn’t exist
Both returned status: success. 🟢🟢
So we built ATO: receipts