The trust loop
for autonomous agents.

Autonomous agents are executing production work around the clock: shipping code, managing infrastructure, handling customer requests. The stack that governs them has not kept up. Today, most organizations track agent liveness with cron jobs and Slack threads. They prove compliance with Notion runbooks handed to auditors. They scope authorization with environment variables and hope.

That stack fails silently and proves nothing.

Dragons closes the trust loop. Every agent gets cryptographic identity through a manifest hash. Every action gets a signed lease that scopes what the agent may do and until when. Every outcome gets a hash-chained receipt in the WorkLedger. Every heartbeat proves the agent is alive and progressing. The result is a replayable, auditable evidence chain from goal to action to outcome — without requiring trust in the operator who ran it.

This is not monitoring. Monitoring watches and reports. Dragons governs and proves. The difference is structural: a monitor tells you what happened; a control plane determines what is permitted to happen and provides cryptographic evidence that it did.

Dragons is not a dashboard. Dashboards display metrics. Dragons issues leases, enforces organism-state lifecycle, and produces signed evidence packs. The surfaces that display Dragons data are views into the control plane — they are not the product.

Dragons is not an orchestration framework. Orchestration decides the order in which tasks execute. Dragons does not schedule, sequence, or route work. It governs what an agent is authorized to do and proves it did it. Temporal, Prefect, and LangGraph orchestrate pipelines; Dragons governs the agents that run inside them.

Dragons is not an observability tool. Observability collects metrics, traces, and logs for human operators to query after the fact. Dragons produces evidence at execution time — hash-chained, signed, replayable without operator trust. The distinction: observability is forensic; Dragons is constitutional.

Dragons is a control plane. The layer between autonomous execution and real consequences. It governs what an agent is, what it may do, whether it is still alive, and whether the work it did can be proven to a third party. One tool, one job.

A governed agent is a dragon. Not the fantasy kind — the infrastructure kind. An autonomous process that runs hot, does real work in production, and is kept alive by a supervisor that knows its identity, its authorization scope, and its liveness state. When it degrades, the supervisor restarts it. When it violates its lease, the supervisor quarantines it. When it completes work, the evidence is signed and chained.

The name holds a deliberate tension. "Dragon" is evocative — it suggests power, autonomy, a thing that operates on its own terms. But the product is precise systems discipline: cryptographic identity, signed leases, organism-state lifecycle, replayable evidence. The brand inhabits the space between the mythic name and the infrastructure function. The name is evocative; the mechanism is exact.

The mark

The pixel-flame dragon is 42 rectangles on a 32x32 grid. It is a flying dragon and a flame at once — the body is flame-orange, the wing is ember-red poking upper-right, the midsection is bright, and the heat trail dives to the lower-left. Every visible element is a deliberate, sized square. No bezier curves, no anti-aliased silhouettes. The mark is a demoscene-lineage pixel-art construction: the pixel as the unit of visual identity.

The thermodynamic spectrum

The color system is borrowed from thermal imaging instruments. High energy is hot: ember, flame, heat. Resolved structure is cool: crystal, blue. The palette encodes physical state, not aesthetic preference. A running agent is cool (stable, resolved). An autohealing agent is flame-pulsing (active hot work). A degraded agent is heat (warm but unstable). A dead agent is ember (the deepest red). The colors do not decorate — they report thermodynamic state.

Six references define the design space Dragons inhabits. Each produces a law that governs every surface, every sentence, every interaction.

Dragons is built by MOGOS Collective — the federation that builds the auditable labor substrate. MOGOS proves that autonomous systems deliver accepted outcomes at lower cost than human teams, with replayable receipts. Dragons is the control plane that makes this proof possible: without governed identity, scoped authorization, and hash-chained evidence, the claim of autonomous delivery is unverifiable.

Every system in the MOGOS federation has a single job. Dragons governs agents. YSO observes. KAIROS schedules. RELOS records receipts. GRANDOS governs goals. They compose — they do not overlap.