API + MCP

One vocabulary and standard for the ecosystem

API v2 exposes complete bilingual standards, controlled vocabulary, and settlement draft contracts. MCP lets agents read and draft while publication, approval, equity writes, and payment remain human-authorized.

View OpenAPI Read standards

MCP protocol2025-11-25

sha256:5098760caa5105bf

Public standards API

Anonymous reads cover complete bilingual standards, terms, TaskSpecs, roles, policies, templates, and brand provenance. Formal writes require real-time TableUser grants and fail closed.

GET /api/v2/manifest

MCP Streamable HTTP

Protocol 2025-11-25. Resources are public; validation and draft tools require an ecosystem API key.

POST /mcp

Callable resources

opc://manifest/current
opc://glossary/current
opc://standards/{sectionId}
opc://taskspec/{taskCode}
opc://roles/{role}
opc://policy/{policyId}
opc://brand/current
opc://split-templates/{version}
opc://organizations/{id}
opc://split-projects/{id}
opc://settlement-runs/{id}
opc://equity-ledgers/{legalEntityId}

Validation and draft tools

resolve_term
validate_standard_reference
validate_taskspec
compare_standard_versions
draft_alignment_report
recommend_role
draft_onboarding_profile
assess_readiness
propose_task_path
simulate_operating_plan
draft_split_project
validate_split_rule
calculate_draft_settlement
explain_settlement
compare_rule_versions
draft_invitation

Reliability boundaries

AI output remains a proposal until human confirmation.
Payment, credentials, split changes, and final submission are not autonomous tools.
MCP cannot publish rules, override FX, approve ledgers, or post equity events.
Tool keys are stored only as hashes; logs exclude secrets and sensitive source text.