#tokenjam
5 posts
-
Claude Code OTel Telemetry: What Cursor and /cost Won't Show
Claude Code emits real OpenTelemetry spans; Cursor and /cost don't. See what the OTel wire exposes and the failure modes the built-in views miss.
-
LangSmith Cost in 2026: Real TCO vs Self-Hosted Alternatives
LangSmith's $39/seat sticker runs ~10.7x that in real TCO. A sourced teardown vs Langfuse self-host and a local-first DuckDB alternative, with real numbers and config.
-
The taxonomy of agent failure: 13 named alerts beat 'anomaly detected' at 2am
Every AI observability vendor ships 'anomaly detected.' That's the wrong abstraction for autonomous agents. Here's the typed vocabulary we ship instead. 13 named failure modes, each with its own trigger, payload, and prescribed response.
-
How to Monitor Claude Code with OTel (Before a $1,700 Bill)
Monitor Claude Code on your laptop in 5 steps: enable Anthropic's OTel telemetry, store spans locally, and alert on retry loops while the agent still runs.
-
AI Agent Drift Detection: Catch It Before Your Rules Decay
AI agent drift detection with no embedding model: Z-scores on tokens, duration, and tool counts plus Jaccard on tool sequences, run over your own sessions.