{"data":{"slug":"netis-heron","name":"heron","tagline":"Performance monitoring tool for LLM APIs and AI agents","github_url":"https://github.com/Netis/heron","owner":"Netis","repo":"heron","owner_avatar_url":"https://avatars.githubusercontent.com/u/5688247?v=4","primary_language":"Rust","stars":101,"forks":10,"topics":["agentic-ai","ai-agent-development","ai-observability","libpcap","litellm","llm-monitoring","llm-observability","llmops","observability","ollama","packet-capture","rust","sglang","vllm"],"archived":false,"github_pushed_at":"2026-08-18T08:24:47+00:00","maintenance_label":"Active","stars_delta_30d":27,"url":"https://www.graphcanon.com/tools/netis-heron","markdown_url":"https://www.graphcanon.com/tools/netis-heron.md","api_url":"https://www.graphcanon.com/api/graphcanon/tools/netis-heron","graph_url":"https://www.graphcanon.com/api/graphcanon/graph?tool=netis-heron","description":"Agent and LLM API performance monitoring via network packet probe. Measures performance of OpenClaw, Claude, Codex, DeepAgents and more — deployed on the provider side, no SDK changes required.","homepage_url":"https://heron-ai.pages.dev","license":"Apache-2.0","open_issues":3,"watchers":2,"ai_summary":"Heron is an open-source performance monitoring utility that captures network traffic to measure the performance of multiple LLMs and AI agents. It deploys on the provider side without requiring changes to SDK.","readme_excerpt":"## 30-Second Quick Start\n\nNo live capture needed. No privileges. Just a `.pcap` with LLM traffic.\n\n```bash\n\n---\n\n# Install (Linux/macOS, user-local, no sudo)\ncurl -fsSL https://raw.githubusercontent.com/Netis/heron/main/install.sh \\\n  | INSTALL_DIR=\"$HOME/.local\" sh\n\n---\n\n## Install & Verify with an AI Agent\n\nRunning an AI coding agent (Claude Code, Codex, etc.)? Hand it the prompt below\nand let it do the install + smoke test. It needs only shell access to the target\nmachine.\n\n```text\nInstall and smoke-test Heron (https://github.com/Netis/heron) on this machine:\n\n1. Read the README and docs/install.md to pick the right install path.\n   Use the one-line installer; user-local (no sudo) is fine.\n2. Verify the binary: `heron --version` and `heron --help` both work.\n3. Smoke-test WITHOUT live capture (no privileges needed): find or fetch a\n   small .pcap with LLM traffic (the repo's testdata/pcaps/ has fixtures),\n   then run `heron --pcap-file <file> --no-retention`.\n4. Confirm the API is up: `curl -s http://localhost:3000/api/health` returns\n   healthy, and `curl -s 'http://localhost:3000/api/traces?limit=5'` returns\n   reconstructed traces.\n5. (Optional, needs CAP_NET_RAW) for a live test: setcap the binary and run\n   `heron -i <iface>`, generate some LLM traffic through the host, then\n   re-check the console at http://localhost:3000.\n\nReport the console URL and the trace count you saw. Don't hard-code or commit\nany host/credential — this repo rejects infra leakage in CI.\n```\n\nThe last line matters: a `check-leakage.sh` CI gate fails any PR that commits a\nprivate IP, plaintext credential, or key — keep your own infra out of anything\nyou push back.\n\n---","github_created_at":"2026-04-08T01:18:41+00:00","created_at":"2026-07-15T10:41:47.894314+00:00","updated_at":"2026-09-20T04:25:36.199581+00:00","categories":[{"slug":"evaluation-observability","name":"Evaluation & Observability","url":"https://www.graphcanon.com/categories/evaluation-observability","markdown_url":"https://www.graphcanon.com/categories/evaluation-observability.md","api_url":"https://www.graphcanon.com/api/graphcanon/categories/evaluation-observability"}],"tags":[{"slug":"agentic-ai","name":"agentic-ai"},{"slug":"ai-agent-development","name":"ai-agent-development"},{"slug":"libpcap","name":"libpcap"},{"slug":"llm-monitoring","name":"llm-monitoring"},{"slug":"rust","name":"rust"}],"trust":{"provenance":{"is_fork":false,"github_id":1204399383,"owner_type":"Organization","methodology":"github_public_v1","parent_repo":null,"near_duplicate_slugs":[]},"computed_at":"2026-09-11T06:00:13.593Z","maintenance":{"label":"Active","score":82,"methodology":"github_public_v1","releases_90d":6,"days_since_push":23,"last_release_at":"2026-08-17T09:53:16Z","stars_delta_30d":27,"open_issues_delta_30d":0},"security_summary":{"status":"no_lockfile","scanner":null,"low_count":0,"high_count":0,"last_scan_at":"2026-07-15T10:41:49.170Z","medium_count":0,"scan_profile":"none","critical_count":0}},"capability_facts":{"scan":{"source":"repo_scan","observed_at":"2026-09-11T06:00:14.129Z"},"languages":{"value":["rust"],"source":"github.language","observed_at":"2026-09-11T06:00:14.129Z"},"license_spdx":{"value":"Apache-2.0","source":"github.license","observed_at":"2026-09-11T06:00:14.129Z"}},"decision_facts":{"hosting":null,"pricing":null,"requirements":null,"constraints":null,"when_to_use":["When you need a provider-side solution that does not require altering existing codebases or SDKs to monitor performance metrics.","For developers looking to avoid complex installations, Heron can be easily set up using pre-defined scripts.","If your deployment requires capturing network packets for performance analysis specifically for LLM APIs and AI agents like OpenClaw and Codex."],"when_not_to_use":["When the need is for an in-agent monitoring tool rather than a network packet-based solution, as Heron operates on traffic.","In environments where live capture requires administrative privileges that are not available to the user performing the installation.","For real-time performance insights without prior deployment because Heron involves a setup phase and typically uses pre-collected `.pcap` files."],"source":"enrich:decision_facts","observed_at":"2026-07-16T21:24:39.603Z"},"constraint_facets":null,"decision_summary":[{"label":"Adopt for","value":"An open-source network traffic analysis tool for monitoring the performance of LLMs and AI agents without requiring SDK changes."},{"label":"License detail","value":"Apache-2.0"}]}}