Agent Capability Directory
The discoverable index of Bossman human and agent-access surfaces.
Human Routes
/agent-capabilities
Agent API
/api/agent/capabilities, /api/agent/capabilities/{module}
Boundary
Read-only in Milestone 0. Changes ship through reviewed code and tests.
Core Registry
Accepted truth for domains, sites, repositories, owners, standards, runners, and decisions.
Human Routes
/core-registry, /domains, /sites, /proposals
Agent API
/api/agent/capabilities/core_registry, /api/agent/core-registry, /api/agent/domains, /api/agent/sites, /api/agent/proposals
Boundary
Milestone 0 shell only. Future writes must pass through audited approvals.
Imports
Registered source connectors, import runs, observations, and pending proposals before accepted truth changes.
Agent API
/api/agent/imports, /api/agent/imports/{source}
Boundary
Import execution writes staging records only. Accepted truth promotion is a separate reviewed path.
SEO Evidence
SEO, analytics, standards, and monitoring evidence imported from the SEO Champion archive and newly collected by Bossman.
Human Routes
/core-registry, /site-evidence
Agent API
/api/agent/capabilities/seo_evidence, /api/agent/site-evidence
Boundary
Evidence imports are append-oriented. Evidence does not become accepted truth automatically.
Standards
Fleet standards, rule definitions, evidence requirements, and compliance state.
Human Routes
/core-registry
Agent API
/api/agent/capabilities/standards
Boundary
Standards are versioned. Site compliance can fail when standards change.
Worker Runs
Observable job runs for imports, checks, standards, integrations, and agent tasks.
Agent API
/api/agent/capabilities/worker_runs
Boundary
Horizon is the queue visibility layer. Run-triggering APIs will be added behind approval scopes.