Contracts¶
Every inter-component boundary in AgentOps is a contract — a versioned, validatable interchange format. These are the interchange files used between skills, the runtime, and external integrations.
-
Repo-local bootstrap, validation, tracker, and done-criteria contract for autonomous orchestration.
-
Repo-local operational contract for bounded autonomous development.
-
RPI run registry specification.
-
Contract for
.agents/rpi/next-work.jsonl. -
Canonical intake-ledger contract for reusable findings.
-
V2 promotion ladder, executable constraint index, and lifecycle rules.
-
Canonical event mapping across Claude, Codex, and manual runtimes.
-
Required flags, tool allowlists, and timeout strategy for non-interactive Claude/Codex execution.
-
Source of truth for Codex runtime skill structure, frontmatter, discovery paths, and multi-agent primitives.
-
Interface contract for adaptive context assembly and token budgeting.
-
Session Intelligence Trust Model
Artifact eligibility contract for runtime context assembly.
-
Process model, locking, keep-awake, and artifact floor for private overnight runs.
-
Canonical
summary.jsonandsummary.mdschema for Dream outputs. -
Structured template for agent scope-escape reporting.
-
Standard references for agent dispatch prompts.
-
Permissive shape covering all historical swarm result files.