OTel Plugin for Hermes Agent
Open your AI agents' black box. The waste is between your agents, not inside them. Four deterministic detectors over one trace file, priced from real usage fields. Hosted at boxdawn.com.
Find where AI apps break first — and block regressions before they ship.
Does your LLM telemetry tell the truth? Replays provider responses through LLM instrumentation libraries and checks every reported value against the wire.
Production-inspired AI observability lab: a traced and evaluated fictional banking RAG assistant with OpenTelemetry, Phoenix, prompt experiments, guardrails, and optional Dynatrace fan-out.
A truth-first visual observatory for understanding, replaying, comparing, and monitoring AI agent runs.
Catch the step where an AI agent contradicts its own earlier decision. CLI and adapters for LangGraph, CrewAI, Letta, OpenInference, DBOS and multi-agent logs, plus an expiry read for long-running agents.
Test whether agent runs preserve the execution data you may need later.
Turn production LLM failures into regression tests, automatically.
🛡️ Catch unsafe AI agent actions before they execute — explainable risk scoring, injection detection, and real-time containment powered by SigNoz & OpenTelemetry.
Recompute what your LLM traces cost, offline, from OTLP, Langfuse or Phoenix exports, with cache and reasoning tokens priced.
Autonomous LLM agent (Qwen2.5-Coder + QLoRA / MLX) that solves Oracle DB administration CTF tasks against a real database, with OpenInference trajectories.
Multi-harness coding-agent tracing with durable SQLite request and session analytics
Production tracing layer for Claude Agent SDK → Langfuse: named traces, session/user injection, usage & cost normalization, redaction, and full-chain CLI OpenTelemetry.
Passive OpenInference/OTLP tracing adapter for Deep Agents Code Hooks v2
OpenTelemetry-based semantic conventions and instrumentation for tracing LLM applications, AI agents, RAG pipelines, tool calls, and evaluations. Python, TypeScript, Java, and Go libraries with context propagation and configurable data masking, compatible with Phoenix, Arize AX, and other OTel backends.
Retry Safety Analyzer: deterministic HIGH/MEDIUM/SAFE when a side-effecting tool may have been retried. Offline import — never executes the trace. HIGH is risk, not proof. External validation open.
Observability and release gating for tool-using AI agents: OpenTelemetry/OpenInference tracing, cost budgets, loop guards, validated tools, and a golden-dataset scorecard that fails CI.
Evidence-backed improvement proposals for AI agents.
To associate your repository with the openinference topic, visit your repo's landing page and select "manage topics."