Validating LLM Tool Call Arguments With Ecto
Ecto schemaless changesets as the pre-execution trust boundary on LLM tool-call arguments, not another output-shape validator.
We can't find the internet
Attempting to reconnect
Something went wrong!
Attempting to reconnect
Writing · Tag
34 posts on security, page 2 of 3. Or browse the full writing index →
Ecto schemaless changesets as the pre-execution trust boundary on LLM tool-call arguments, not another output-shape validator.
An append-only Ecto schema for LLM calls, the fields an auditor actually samples, and why provider request IDs are what makes an entry defensible.
The two documents an enterprise reviewer wants for your AI agent, the allow-list and the tool-call log, specified field by field and generated from code.
Security checklists only ever grow, because nobody who writes them pays the cost of a control a small team cannot staff. Which to defer, and which never wait.
The compile-time trap in config.exs, why System.fetch_env! should crash your boot, and how to actually wire Fly secrets and LLM keys through a Phoenix release.
Tenant isolation for AI SaaS in Phoenix: per-tenant API key tradeoffs, Ecto context scoping, and the pgvector bug that leaks one tenant into another.
José Valim says Anthropic hasn't separated its security case from its commercial interest. He's right, and it's every frontier lab's problem.
Microsoft just made vulnerability detection cheaper and more accurate. For a 12-40 person AI startup, that was never the bottleneck.
Microsoft's security system hit 96% on a benchmark this week. Here's what that score actually measures, and how to read any vendor's benchmark claim.
The enterprise shadow-AI playbook assumes a CASB budget you don't have. Here's the free, no-blame version for a 5-15 person team: three steps, zero procurement.
A security engineer builds and a vCISO decides. Which one your startup needs first, stage by stage, and the cases where the answer is neither.
Your agent's tools need your production keys; the agent itself never should. The broker pattern, per-tool scoping, and what a five-person team can skip.