i
Class-conditional calibration
A coding agent and a research agent are not held to the same statistics. UNITARES learns separate baselines per agent class from production telemetry, so drift in one class is not masked by noise from another.
— A research preface, with running code
Runtime governance
for heterogeneous
AI-agent fleets.
Most AI safety reasons about agents in two windows: pre-deployment evaluation and post-incident forensics. Between them — while agents are actually running — there is a measurement gap. Continuous, class-calibrated self-state telemetry, with signed provenance behind every intervention, fills it.
$ git clone CIRWEL/unitares && docker compose up → http://localhost:8767/mcp/
— The receipts
- Paper
- UNITARES: Information-Theoretic Governance of Heterogeneous Agent Fleets · Wang, 2026
- DOI
- 10.5281/zenodo.19647159 (concept · resolves to latest)
- Author
- Kenny Wang · Independent Researcher · ORCID 0009-0006-7544-2374
- Patents
- 9+ provisional, filed
- Tests
- 6,200+ at 77% coverage
- Production
- Continuous since November 2025
- Code
- github.com/CIRWEL/unitares · server
github.com/CIRWEL/unitares-governance-plugin · Claude Code / Codex client
§01 — The thesis
Runtime self-state is the missing layer.
The discipline of AI safety has built two strong instruments and one large gap. Pre-deployment evaluation tells you whether a model behaves on a benchmark. Post-incident forensics tells you what went wrong after it didn't. In between — across the hours, days, and weeks an agent is actually running — we have largely been guessing.
Logs are what an agent did.
Self-state is what it was while doing it.
CIRWEL Systems builds the runtime layer that closes the gap. Each agent carries a continuous, four-dimensional self-state vector. The vector is calibrated against a baseline specific to the agent's class, because a long-running coding assistant does not behave like an ephemeral parser, and neither behaves like an embodied service. Drift is detected against the right reference, not an averaged one. Every governance verdict — proceed, guide, pause, reject — carries a signed lineage back to the observation that produced it.
The framework is described in a paper, covered by nine provisional patents, and has been governing CIRWEL's own development fleet without interruption since November 2025.
§02 — Three pillars
ii
Drift detection at runtime
Continuous state observation catches behavioral drift while it is happening, not in the post-incident review. Verdicts arrive early enough to intervene, late enough to be evidence-based.
iii
Auditable provenance
Every intervention carries a signed lineage back to the observation that triggered it. Regulators, underwriters, and the next-shift human can replay the chain — not just read a verdict.
§03 — In production
Governing its own development.
The system you read about on this page also wrote, tested, and shipped a meaningful fraction of itself. CIRWEL's development fleet — a heterogeneous mix of long-running resident agents, short-lived coding sessions, an embodied edge service, and a Discord bridge — has been governed continuously by UNITARES since November 2025.
Living under one's own framework is the cheapest credibility a research operator can offer. We treat it as the floor, not the ceiling.
This page is part of the loop. The colophon below shows the exact commit and build time that produced what you are reading.
§04 — Engage
Three ways in.
- Read The paper: UNITARES (Wang, 2026). CC-BY 4.0, citable via the DOI above.
- Run The server at CIRWEL/unitares boots locally with one docker compose up. The plugin at CIRWEL/unitares-governance-plugin drops into Claude Code or Codex and emits real verdicts against it. Both are MIT.
- Write Direct to the founder: founder@cirwel.org. Funders, researchers, and integrators all welcome.