Deschide articolul original într-un tab nou Schimbarea acestei opțiuni se aplică imediat și reîncarcă pagina.
LECTURA
Înapoi la articole →

Teng Li · ML Engineer

tengli.dev/feed.xml
Vizitează site-ul

Cel mai recent articol

Ultima verificare

Urmăritori 0 urmăritori
Articole: 4

  1. My eval said a perfect MCP server was broken. It was the eval that was lying.

    (Teng Li · ML Engineer)

    Three rounds of calibration, four real servers, $0.60 in API costs — how I made an LLM-powered tool-selection benchmark fair enough to publish, and what it revealed about static lint rules predicting live model behavior.

  2. I lint-scanned 36 popular MCP servers. A third of them are failing your agent.

    (Teng Li · ML Engineer)

    Your MCP server can be 100% spec-compliant and still be unusable by an agent.

  3. Why this blog exists

    (Teng Li · ML Engineer)

    Production AI agents fail in ways that never show up in demos.

  4. Designing tools LLMs can actually use: five failure modes I keep seeing

    (Teng Li · ML Engineer)

    Your tool API can be perfectly engineered and still confuse every model that touches it.