Keydris documentation
How Keydris works, in dependency order.
The landing page teaches why Keydris exists. These docs teach how it works, in the order the concepts depend on each other. Read top to bottom the first time; every page also stands alone.
The learning path
(01)IntroductionWhat Keydris is, the mental model, and what it deliberately does not do.(02)QuickstartFour terminal lines to your first governed session.(03)Identity vs authorityThe core concept: why knowing which agent is calling is not enough.(04)KITs & devicesThe short-lived session identity an agent presents, and what it binds.(05)The execution boundaryThe broker, the KIT Reader, and exactly what crosses to the control plane.(06)PoliciesAuthoring the rules: decisions, defaults, conditions, and versioning.(07)Audit evidenceDecisions and outcomes as distinct, reviewable, exportable facts.Then, by need
(08)MCPGoverning MCP tool calls and resource reads, in both modes.(09)GitHubGoverning repository actions without handing the agent your access.(10)SlackGoverning what agents may read and post in selected public channels.(11)Local commandsSupported shell commands checked before the harness executes them.(12)Console & operationsThe workflows an operator owns: agents, policies, evidence, members.(13)CLI & vocabularyThe verified command surface and the exact meaning of each term.Vocabulary
- Agent
- an operator-managed identity for an AI workload.
- Device
- an enrolled installation on which an agent runs.
- KIT
- the short-lived runtime identity for one agent session.
- Broker
- the enforcement point in your environment, on the agent's side.
- Reader
- optional enforcement on your MCP servers, on the receiving side.
- Policy
- operator-authored rules governing an agent's actions.
- Decision
- what Keydris concluded about one governed action: ALLOW, REJECT, or APPROVAL REQUIRED. Exactly these three.
- Reason
- why a decision came out that way (for example scope mismatch, or authority verified). A reason travels under a decision; it is never the decision.
- Outcome
- what happened when an allowed operation was attempted, recorded separately from the decision.
- Governed
- resources you enroll + the policy you assign, for supported actions.