ControlSurface v0.1.0-rc.1

[DevChiniwala]

ControlSurface v0.1.0-rc.1

First public release candidate of ControlSurface, an open-source production engineering platform for AI agents.

Highlights

• OTLP-native agent tracing with source spans and AgentRunGraph projections.
• Production Health and AI-specific SLO evaluation.
• Deterministic structural failure clustering.
• Incidents with inspectable Change Ledger evidence.
• Production-derived, human-reviewed regression cases.
• Paired candidate evaluation.
• Reproducible release gates with content-addressed evidence.

Run it locally

Follow the Quickstart to start the supported Docker Compose deployment, create the owner, and send the first trace.

The refund-agent lifecycle demonstrates healthy traffic, a breaking tool-schema update, failure clustering, incident evidence, regression review, a fixed candidate, and a passing release gate.

Release status

This is an early source release candidate. Docker Compose is the supported V1 self-hosting path. The trust model is a trusted single owner on a trusted host; see Security and telemetry privacy and Operations.

Public Python-package and container-image distribution are intentionally deferred until the artifact-specific license bundle is complete. The tag includes source archives generated by GitHub; the full real-product demo is attached to the release separately.

Verification

The tagged commit must pass the quality and hero-e2e GitHub Actions checks and a clean-volume Compose boot before this release is published. Detailed evidence is maintained in the release audit.

---

This discussion was created from the release ControlSurface v0.1.0-rc.1.