[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"readme:agentsight":3},"\u003Ch1>AgentSight: System-wide AI agent profiling and monitoring with eBPF\u003C\u002Fh1>\n\u003Cp>\u003Ca href=\"https:\u002F\u002Fopensource.org\u002Flicenses\u002FMIT\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-green.svg\" alt=\"License: MIT\" \u002F>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Feunomia-bpf\u002Fagentsight\u002Factions\u002Fworkflows\u002Fci.yml\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fgithub.com\u002Feunomia-bpf\u002Fagentsight\u002Factions\u002Fworkflows\u002Fci.yml\u002Fbadge.svg\" alt=\"CI\" \u002F>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2508.02736\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FarXiv-2508.02736-b31b1b.svg\" alt=\"arXiv:2508.02736\" \u002F>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fdl.acm.org\u002Fdoi\u002F10.1145\u002F3766882.3767169\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDOI-10.1145%2F3766882.3767169-blue.svg\" alt=\"DOI:10.1145\u002F3766882.3767169\" \u002F>\u003C\u002Fa>\u003C\u002Fp>\n\u003Cp>\u003Cstrong>English\u003C\u002Fstrong> | \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Feunomia-bpf\u002Fagentsight\u002Fblob\u002Fmaster\u002FREADME.zh-CN.md\" rel=\"nofollow ugc noopener\">中文\u003C\u002Fa>\u003C\u002Fp>\n\u003Cp>AI agents can run commands, rewrite files, spawn processes, and contact remote services—but when a run fails, stalls, or behaves unexpectedly, their logs rarely tell the whole story.\u003C\u002Fp>\n\u003Cp>AgentSight is a local-first \u003Ccode>top\u003C\u002Fcode>\u002F\u003Ccode>strace\u003C\u002Fcode>-like observability tool for AI agents. It connects prompts, model calls, and tool decisions to their real effects on your machine.\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>See what the agent is doing now.\u003C\u002Fstrong> Monitor active sessions, processes, resource, model and tool calls, file and network activity.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Understand failures and improve agent behavior.\u003C\u002Fstrong> Connect prompts and skills to errors, then turn them into better instructions and skills.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Find where time, tokens, and resources go.\u003C\u002Fstrong> Spot slow steps, retry loops, repeated model or tool calls, token-heavy sessions, and resource-hungry processes.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Audit data movement and security-sensitive effects.\u003C\u002Fstrong> See which services received requests, which files changed, and whether agents stayed and give safety\u002Fsecurity suggestions.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Use it with your existing agents.\u003C\u002Fstrong> Run AgentSight around Claude Code, Codex, Gemini CLI, OpenCode, OpenClaw, or any command—without an SDK, proxy, or vendor integration.\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>No SDK, no proxy, no vendor integration. AgentSight observes with eBPF and TLS traffic tracing, so it works even when the agent is a\nclosed-source CLI. \u003Cstrong>✨ Zero SDK Required\u003C\u002Fstrong>\u003C\u002Fp>\n\u003Ch2>Quick Start\u003C\u002Fh2>\n\u003Cpre>\u003Ccode class=\"language-bash\">cargo install agentsight\n# or: wget https:\u002F\u002Fgithub.com\u002Feunomia-bpf\u002Fagentsight\u002Freleases\u002Flatest\u002Fdownload\u002Fagentsight && chmod +x agentsight\n\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Cdiv align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fgithub.com\u002Feunomia-bpf\u002Fagentsight\u002Fraw\u002Fmaster\u002Fdocs\u002Ftop-mode-demo.png\" alt=\"AgentSight top live session view\" width=\"1000\" \u002F>\n  \u003Cp>\u003Cem>Live sessions ranked by model, session tokens, health, process family, tool calls, file activity, and network activity\u003C\u002Fem>\u003C\u002Fp>\n\u003C\u002Fdiv>\u003Cpre>\u003Ccode class=\"language-sh\">agentsight top\n\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Cdiv align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fgithub.com\u002Feunomia-bpf\u002Fagentsight\u002Fraw\u002Fmaster\u002Fext\u002Fvis\u002Fexamples\u002Factplane-agent-nebula.gif\" alt=\"Agent Nebula replay of Agent development across the ACTplane repository\" width=\"1000\" \u002F>\n  \u003Cp>\u003Cem>Agent Nebula replays how coding agents read, write, create, rename, and delete files across the ACTplane repository\u003C\u002Fem>\u003C\u002Fp>\n\u003C\u002Fdiv>\u003Cpre>\u003Ccode class=\"language-sh\">agentsight vis\n\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Cdiv align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fgithub.com\u002Feunomia-bpf\u002Fagentsight\u002Fraw\u002Fmaster\u002Fdocs\u002Fflamegraph-example\u002Fsemantic-flamegraph-top200.svg\" alt=\"Semantic flamegraph of the top 200 agent stacks\" width=\"1000\" \u002F>\n  \u003Cp>\u003Cem>Width is system-effect weight; the uneven stack height shows prompt, tool-call, process, and effect paths ending at different depths. See the \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Feunomia-bpf\u002Fagentsight\u002Fblob\u002FHEAD\u002Fdocs\u002Fagentpprof.md#example-flamegraphs\" rel=\"nofollow ugc noopener\">agentpprof guide\u003C\u002Fa> for the other profiles and how widths and stack depths are drawn.\u003C\u002Fem>\u003C\u002Fp>\n\u003C\u002Fdiv>\u003Ch2>🚀 Why AgentSight?\u003C\u002Fh2>\n\u003Ch3>Traditional Observability vs. System-Level Monitoring\u003C\u002Fh3>\n\u003Cp>Application-level tools such as \u003Ca href=\"https:\u002F\u002Fdocs.langchain.com\u002Flangsmith\u002Fobservability-concepts\" rel=\"nofollow ugc noopener\">LangSmith\u003C\u002Fa>, \u003Ca href=\"https:\u002F\u002Flangfuse.com\u002Fdocs\u002Fobservability\u002Foverview\" rel=\"nofollow ugc noopener\">Langfuse\u003C\u002Fa>, and \u003Ca href=\"https:\u002F\u002Farize.com\u002Fdocs\u002Fphoenix\u002F\" rel=\"nofollow ugc noopener\">Phoenix\u003C\u002Fa> are great for traces, prompts, tokens, evals, and latency when you own the application code. Gateway\u002Fproxy tools such as \u003Ca href=\"https:\u002F\u002Fdocs.helicone.ai\u002Fgetting-started\u002Fintegration-method\u002Fgateway\" rel=\"nofollow ugc noopener\">Helicone\u003C\u002Fa> are useful when you can route provider traffic through a managed endpoint.\u003C\u002Fp>\n",1788652552323]