CI Agent Telemetry in Beacon | Asymptote
What is CI agent telemetry?
CI telemetry started with Claude Code activity captured to Beacon JSONL, GitHub Actions artifacts, and customer-managed export destinations, then expanded to multi-step CI workflows.
Collection path: Temporary local collector through beacon ci
Authoritative date: June 7, 2026 from v0.0.45 in the Beacon CLI changelog.
How Beacon reads CI agent telemetry
beacon ci exec or beacon ci start and beacon ci finish run an ephemeral collector, capture supported agent telemetry during the job, validate it, and retain the runtime JSONL as an artifact or upload target.
The result is normalized into Beacon's endpoint event schema so security teams can investigate activity by harness, repository, session, user, model, command, file, tool, MCP server, or approval context where the runtime exposes those fields.
Telemetry coverage
- Prompt, tool, command, and file events where emitted by supported agents
- GitHub Actions run, repository, workflow, branch, and PR context
- Runtime JSONL artifacts
- S3, GCS, and SIEM handoff paths
What you can do now
- Capture AI agent activity in CI without a long-running endpoint service.
- Fail automation with scan gates when detections cross severity thresholds.
- Upload completed runtime logs to object storage or a SIEM.
Get started
- Wrap the agent command with beacon ci exec.
- Use beacon ci start and beacon ci finish for multi-step workflows.
- Publish the JSONL artifact or configure an upload destination.