opentel-mcp brings OpenTelemetry silent-failure detection to Python MCP servers
Thirumalaiboobathi · reddit · 2026-10-27
Developer Thirumalaiboobathi released opentel-mcp, an OpenTelemetry instrumentation library for Python MCP servers that catches "silent failures" invisible to standard tracing: MCP tools often return isError: true inside a successful response, so normal OTel records a green span while the agent actually received permissiondenied or notfound.
- Fingerprints and categorizes failures (validation, timeout, network, auth, dependency, serialization, internal) even when IDs/paths/timestamps differ
- Tracks calls, errors, silent failures, and duration
- Scrubs secrets (Bearer tokens, API keys, JWTs); never captures tool arguments or results
- Never raises into your server; ships a zero-backend dev dashboard via npx opentel-mcp-ui
Compared with the mcp 2.x SDK's built-in tracing (marks errors but no metrics/fingerprints/categories, and logs raw exceptions with stack traces that can leak secrets) and Traceloop/OpenInference (no isError detection; Traceloop captures inputs/outputs). It's a port of an established JS package with 7,000+ npm downloads; works with mcp SDK 1.25–2.x, tested across Python 3.10–3.14 with mypy --strict.
More from coding & agent
- DeepSeek Harness plugin ecosystem is booming as cheap, mid-tier-but-solid model draws tinkerers — op7418 · 2026-10-27
- Open-source auto-research agent hits Top 2 on GPU kernel challenge in 7 hours — sguada · 2026-10-27
- 'Codex decided': devs admit making up reasons in code review for agent-written code — FlatStill2540 · 2026-10-27
- Running AI agents on a legacy IBM WMS: enforce tool order, script checks, verify evidence — ksrijith · 2026-10-27
- 365 public MCP servers tested: 51% lack tool input descriptions, the most common bug — abhiramreddykalluru · 2026-10-27
- 300 Agents, 60 People: Runlayer Found Engineer Details Real-World Agent Security Failures — AI Engineer · 2026-10-27