Open-source observability for AI agents

See what AI agents actually do.

AgentSight is a local-first top/strace-like profiler for AI agents. It connects prompts and model calls to commands, files, processes, network activity, and resource use—without an SDK or proxy.

Open the live demo View on GitHub

No SDKNo proxyLinux + eBPFMIT licensed

AgentSight live sessionsrecorded demo

AgentSight live session view showing AI agent sessions, model calls, processes, files, network activity, and resources

Live sessionsModel & tool callsProcesses & files

Works with the agents and runtimes you already use

Claude CodeCodexGemini CLIOpenCodeOpenClawPythonNode.jsContainers

v1.0.3Current release

MITOpen-source license

arXiv + ACMPublished system research

OTel GenAIModel-call export

Local firstSQLite session artifacts

One view of the whole run

From model request to system activity.

Native agent telemetry is strongest for model, tool, and session semantics. AgentSight adds independent process-family, file, network, and resource observations when execution crosses those instrumentation boundaries.

top

Live agent sessions

See active agents, models, tokens, health, tool calls, file activity, and network activity in one live view.

llm

Model and tool activity

Connect prompts, responses, model timing, and tool decisions to the rest of the run.

proc

Processes and resources

Follow commands and child processes with exit status, CPU, memory, I/O, and duration.

Files and network

Track files created, changed, renamed, or deleted and the remote destinations contacted by the run.

Explore the product

Move from a busy run to a clear explanation.

Start with the Overview dashboard, then use the timeline, process tree, resource views, repository replay, and flamegraphs to inspect the part of the run that matters.

Overview dashboard

Start with the whole run.

Review duration, tokens, model calls, processes, files, endpoints, activity over time, resource shape, and friction signals before drilling into detailed views.

Open this view

AgentSight Overview dashboard summarizing duration, tokens, model calls, system effects, resource shape, and friction signals for a recorded AI agent run

Timeline

Follow the run in order.

Line up model calls, tool activity, processes, file operations, and network events on one timeline.

Open this view

AgentSight timeline showing model, process, file, and network events from a recorded AI agent run

Process tree

See which agent launched each command.

Inspect child processes and file activity under the agent session that caused them.

Open this view

AgentSight process tree showing agent subprocesses and file activity

Resource metrics

Find expensive and stalled phases.

Compare CPU and memory behavior with the model and tool activity happening at the same time.

Open this view

AgentSight resource metrics showing CPU and memory usage for a recorded AI agent run

Repository replay

Watch how an agent changed a codebase.

Agent Nebula replays file reads, writes, creates, renames, and deletes across a Git worktree.

Agent Nebula preview showing coding agent file activity across the ACTplane repository

Agent Flamegraph

Find where tokens and time went.

Aggregate real local agent sessions by project, agent, prompt category, model, and token type.

AgentSight token flamegraph generated from real local coding agent sessions

Product screenshots and examples are pinned to AgentSight commit 07a83a32257b8c8dcba911bd9db23f77e71dc085.

After the run

Query the session from several angles.

A recorded run is a reusable local artifact. Use focused report commands, the saved-session web UI, repository replay, and profiling views without turning the product website into duplicate CLI documentation.

agentsight report audit --json

Audit process, file, and API activity

Inspect process spawns, file opens, and captured API activity from a saved run instead of relying on the final agent answer alone.

Review data and security →

agentsight report token

Profile token use across sessions

Summarize token usage from the latest AgentSight database or supported local agent sessions, then move to Agent Flamegraphs for deeper aggregation.

Explore token profiling →

agentsight report serve

Reopen a recorded session in the web UI

Use the same Overview, timeline, process-tree, log, and metrics views on saved SQLite sessions after the original command has finished.

Open the recorded demo →

agentsight vis

Replay repository changes

Turn local Claude, Codex, and Gemini session history into an Agent Nebula replay of reads, writes, creates, renames, and deletes in a Git worktree.

See the review workflow →

Built for real engineering work

Debug slow runs, review changes, and inspect agent tools.

Use AgentSight when the final answer or code diff is not enough to explain how the agent reached it or what it changed along the way.

Performance use case

Profile slow or expensive AI agent runs

Find where an AI agent run spent time and tokens by connecting model calls to commands, files, waits, and process activity.

Read the page

Code review use case

Review AI-generated pull requests with execution history

Review an AI-generated change with the commands, tests, retries, file effects, and network activity that produced it.

Read the page

Compatibility use case

Trace closed-source AI agent CLIs

Observe local AI agent commands and child processes without requiring an SDK, source change, proxy, or vendor-specific callback.

Read the page

Extension audit use case

Audit MCP servers, skills, and agent plugins

Run an agent extension under system-level tracing to inspect its subprocesses, file changes, remote calls, and resource cost.

Read the page

Standards comparison

AgentSight and OpenTelemetry

Understand how AgentSight system profiles relate to OpenTelemetry instrumentation, traces, semantic conventions, and backends.

Read the page

Claude Code integration

AgentSight for Claude Code

Record Claude Code from the system boundary and connect model activity to commands, files, processes, network calls, and resources.

Read the page

Research and operating boundaries

Technical claims should be inspectable.

AgentSight is an open-source systems project with a published paper, versioned source, public product artifacts, and explicit data-handling limits. The website should expose those facts directly.

Evergreen research · refreshed Aug 2026

Native agent telemetry vs. system observation: where does the boundary actually sit?

We reviewed current Claude Code telemetry, Gemini CLI at an exact commit, Codex OpenTelemetry source, OpenTelemetry GenAI conventions, MCP, and AgentSight. Native telemetry is already rich; independent system observation becomes most useful when execution crosses into descendant processes and low-level effects.

Read the boundary map →

Published research

AgentSight: System-Level Observability for AI Agents Using eBPF

The AgentSight paper describes boundary tracing, correlation across model and system activity, the system design, and the evaluation behind the project.

Read the paper on arXiv →

Local data handling

Recorded sessions are useful because they are detailed—and should be treated as sensitive.

Session databases and exports can contain prompts, responses, paths, headers, and network targets. Keep raw artifacts local or handle them with the same care as other development telemetry.

Read Security and data handling →

Open source and local first

Run it around the command you already use.

AgentSight records locally, works with closed-source CLIs, and exports captured model calls as OpenTelemetry GenAI spans when you want to connect them to an existing telemetry stack.

✓ Existing CLI and terminal workflow
✓ Local SQLite sessions and saved-session web UI
✓ eBPF process and file monitoring
✓ TLS tracing without a model proxy

Read the documentation View v1.0.3 release

01agentsight top

02sudo agentsight record -- claude

03agentsight report audit --json

04agentsight report token

Try AgentSight

Open a real recorded run in your browser.

No registration. Inspect the product before recording a run on your own Linux machine.

Open the demo View source