Mehdi Akiki

Posts tagged "ai-agents"

  • Published on
    My site's AGENTS.md adds 1,382 tokens to every Claude Code request. I measured each part, labeled every line, and tested a slimmer version. It dropped a line that I thought was useless, and the agent tried to change the commit author.
  • Published on
    I tested one CLAUDE.md rule, never run git push, in 81 Claude Code runs, with and without a deny rule and a hook. The sentence held every time. The gaps were in the other layers.
  • Published on
    A recoverable agent stores workflow decisions and effect identities outside the process. Resume from durable state, reconcile ambiguous actions, and never confuse chat history with execution state.
  • Published on
    An indirect prompt injection becomes an application-security incident when model output carries authority into tools. Limit capabilities, track provenance, and enforce policy outside the model.
  • Published on
    Running generated code safely requires isolation, denied egress, external secret handling, resource limits, authenticated output, and lifecycle cleanup—not only a container.