Inspect the team before use
Active database publicationAI Model Governance and Evaluation
Use-case risk, provenance, evaluation and independent trustworthiness evidence joined into an accountable AI governance pack.
Opens the installed OpenCorvus desktop app. The client re-downloads and re-verifies this exact revision, then asks you for an install scope. No desktop app yet? Install OpenCorvus first
Bound execution graph
Workflow and handoffs
The graph shows who owns each step and which earlier result it depends on. It explains the team's collaboration contract; it is not live task progress and does not schedule Agents from the page.
fa7f232c5 nodes- 01ai-use-case-risk-governance-analystMaps use-case risk and governance.
root - 02model-data-provenance-documentation-analystMaps model and data provenance.
root - 03ai-evaluation-design-results-analystMaps evaluation protocol and results.
root - 04ai-independent-trustworthiness-reviewerChallenges trustworthiness evidence.
root - 05ai-model-governance-evaluation-ownerJoins all roots without accepting risk or deploying a model.
← ai-evaluation-design-results-analyst + ai-independent-trustworthiness-reviewer + ai-use-case-risk-governance-analyst + model-data-provenance-documentation-analyst
Intended scope
Use for bounded AI model governance and evaluation evidence.
- Product pillarscode + work
- Agent roster5 declared roles
- Declared workflows1 complete responsibility and dependency graphs
- Full capability projection1 Skills · 3 tools · 0 MCP refs
Roles and ownership
This roster comes from the exact package. Agent identity and base role participate in the real runtime; avatars, ratings, and personas do not influence selection.
- AI Use-Case Risk and Governance Analyst
exploreMaps intended use, affected groups, risk taxonomy, controls, owners and decision authority. - Model and Data Provenance Analyst
delegated-workerVersions model, dataset, prompt, tool, configuration, limitation and documentation evidence. - AI Evaluation Design and Results Analyst
delegated-workerDesigns claim-bound protocols and checks row-level, slice, aggregate and reproducibility evidence. - Independent AI Trustworthiness Reviewer
delegated-workerChallenges robustness, bias, privacy, security, human-oversight and evaluation validity claims. - AI Model Governance and Evaluation Owner
delegated-workerJoins four branches into one claim-risk-evidence-decision register.
Workflow and handoffs
The graph shows who owns each step and which earlier result it depends on. It explains the team's collaboration contract; it is not live task progress and does not schedule Agents from the page.
ai-model-governance-evaluation-review
AI Model Governance and Evaluation Review
Four independent AI evidence branches converge into one accountable review pack.
- 01ai-use-case-risk-governance-analystMaps use-case risk and governance.depends_on: []
- 02model-data-provenance-documentation-analystMaps model and data provenance.depends_on: []
- 03ai-evaluation-design-results-analystMaps evaluation protocol and results.depends_on: []
- 04ai-independent-trustworthiness-reviewerChallenges trustworthiness evidence.depends_on: []
- 05ai-model-governance-evaluation-ownerJoins all roots without accepting risk or deploying a model.depends_on: ai-evaluation-design-results-analyst, ai-independent-trustworthiness-reviewer, ai-use-case-risk-governance-analyst, model-data-provenance-documentation-analyst
Capability scope
The full projection includes built-in, default, and package-owned references. The package-owned subset is disclosed separately and never presented as the complete permission surface.
- Full Skill projection1 unique references; 1 are the package-owned subset
- Full tool projection3 unique references; 0 are the package-owned subset
- Full MCP projection0 server/tool/prompt/resource references; 0 are the package-owned subset
- Configuration0 fields · 0 required
Version and provenance
During publication import, the website reconstructs the canonical file tree and complete display facts from the exact ZIP referenced by the signed catalog. The local Manager validates shape, identity, topology, and digest again during install.
- Identitybuiltin/ai-model-governance-evaluation
- Version2026.08.13.1
- Package SHA-256fa7f232c19480a517d2aceb9306140038d14ea4a00e9230e2c56f7572ee241cb
This record belongs to the active database publication. The importer checked it field by field against the OpenCorvus-signed catalog, exact ZIP, and content digest. The signing private key never reaches the production server; third-party publisher identity and independent review services are not open yet.
Install boundary
The page can hand this exact revision to an installed OpenCorvus client, or stream its content-addressed ZIP. Neither path installs or activates a team from the web: the client re-downloads the archive, re-verifies its bytes, digest, and shape, and requires an explicit scope.
- 01Resolve the exact revision in the website database, verify ZIP bytes and SHA-256, then hand it to the client or start the download response
- 02Recompute the digest and strictly validate the selected package during client import
- 03Require the operator to choose project or user-global scope
- 04Import atomically through the Manager and return a mutation receipt
- 05Select the installed Squad separately before the active profile changes