Our take
Activeregular commits and releases- Activity
- 56
- Community
- 15
- Issues
- 46
- Pull requests
- 46
+18 stars in the last 30 days
- Fewer contributors than 86% of the projects we track
AI Observer is an actively maintained local observability tool with useful OTLP and file-based workflows, but its recent development is heavily concentrated in one contributor.
Good fit if
- You want local visibility into supported AI coding assistants and are comfortable running a Go-based single-binary service or Docker image.
- You need both live OTLP monitoring and local-file import or watch workflows for Claude Code, Codex, or Gemini.
- You want to keep telemetry locally and can configure network access and authentication to suit your environment.
Look elsewhere if
- You need local-file import or watch support for GitHub Copilot or OpenCode, which are OTLP-only here.
- You require an authentication-protected service by default. Authentication is optional and must be configured.
- You need a project with broadly distributed maintenance, since most recent contributions come from one person.
All health signals
| Last commit | 2026-11-06 (today) |
|---|---|
| Commits, last 90 days | 13 |
| Releases, last 12 months | 9 (latest v0.7.0, 2026-11-06) |
| Contributors | 2 (top contributor: 96% of commits) |
| Issues closed, last 90 days | 1 |
| Pull requests merged, last 90 days | 2 |
| Project age | 11 months |
Checked on 2026-11-07 with the GitHub API.
Project history
Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.
Stars
297Forks
28Change since Nov 17, 2026: +1 stars·0 forks
Overview
AI Observer is a self-hosted, single-binary observability backend built on OpenTelemetry and DuckDB. It addresses the gap between using multiple local AI coding assistants and being able to inspect their usage, costs, latency, errors, and session activity in one place.
It supports Claude Code, Gemini CLI, OpenAI Codex CLI, GitHub Copilot, and OpenCode. Telemetry can arrive over OTLP, while local session files provide import and watch options for Claude Code, Codex, and Gemini. Data stays in a local DuckDB database and is presented in a dashboard with live updates.
Key Features
- Ingest OpenTelemetry traces, metrics, and logs over HTTP, including JSON and Protobuf formats.
- View AI assistant activity in a dashboard with real-time updates and configurable, saved dashboards.
- Track token usage and estimated costs across supported tools and model aliases.
- Import historical Claude Code, Codex, and Gemini sessions, or watch their files for new activity without configuring OTLP exporters.
- Search conversations and inspect transcripts, logs, traces, metrics, and costs through read-only MCP tools when enabled.
- Export telemetry to Parquet, with an optional DuckDB views database.
- Redact secrets before storing new telemetry, with an option to redact additional personal data.
Use Cases
- Developers using several coding assistants can compare their activity and estimated costs in one local dashboard.
- Teams investigating slow responses or errors can inspect associated traces, logs, and session activity instead of relying only on provider-specific views.
- Claude Code, Codex, and Gemini users who want historical or live session monitoring without setting up OTLP can use import or watch mode.
- Developers who want an AI agent to search previously captured conversations can enable the MCP endpoint and connect a supported agent.
What you need
Detected in the repository
- Go (main language on GitHub)
- A Dockerfile, so it can run in a container
- Automated checks on GitHub Actions
License in plain words
MITpermissive
- Commercial use: yes
- Modify and redistribute: yes
- You must keep: the copyright and license notice
- Share your changes: no
A summary, not legal advice: the LICENSE file is what applies.
Getting Started
Run the Docker image with local-only port bindings and a persistent data volume:
docker run -d \
-p 127.0.0.1:8080:8080 \
-p 127.0.0.1:4318:4318 \
-v ai-observer-data:/app/data \
--name ai-observer \
tobilg/ai-observer:latest
Open the dashboard at http://localhost:8080. See the README for tool setup, configuration, and other installation options.
Alternatives
- agent-observability: It monitors Claude Code and Codex through Prometheus, Loki, Tempo, and Grafana, rather than using a DuckDB-backed local dashboard with OTLP and file imports.
| Project | Language | License | Stars | Status |
|---|---|---|---|---|
| ai-observer | Go | MIT | 288 | Active |
| agent-observability | Go Template | – | 14 | Quiet |
Considerations
- The project is young, although recent commits and releases indicate active development. Its contributor base is small and highly concentrated, so continuity depends substantially on one maintainer.
- The database is DuckDB, and only one process can open it at a time. Stop the server or watcher before running import, export, or delete commands.
- The service has no authentication by default and may store prompts, code, and tool output. The binary binds to localhost by default; network exposure should be deliberate, and authentication should be configured when appropriate.
- File-based ingestion is limited to Claude Code, Codex, and Gemini. GitHub Copilot and OpenCode require OTLP, and some operational metrics are available only from live telemetry.
- The repository metadata reports no test suite. Consider that when evaluating it for deployments that require tested releases or established operational guarantees.
Found this useful?
Share it with someone who would like ai-observer.