Capability matrix
Live capabilities are available now; any open limitations are listed alongside them. Planned capabilities are unavailable. A capability never satisfies an enterprise control by implication.
Download governed sales/API export (JSON) →| Capability | Edition / status | Limitations | Evidence owner | Review |
|---|---|---|---|---|
Website AI compliance scan CAP-PUBLIC-SCAN Documentation | All plans Live | Requires sign-in, a project and scan permission; there is no anonymous scan. Only shared report links can be opened without signing in, and they expire after 30 days. Public pages only: target-site authenticated scanning is refused. Scans are rate-limited per client address (12 per 10 minutes). Production-like rate-limit, retention and cross-tenant exercises remain pending. | Product and Security Verification record | 2026-11-02 |
Multi-persona governance workspace CAP-GOV-WORKSPACE Documentation | All plans; teammates and client users need Business, Enterprise or Platinum Live | Account owner, company admin, client member and token client portal views are implemented. Token portal clients can respond to findings and evidence requests (with a link to a file) and reply to message threads; acknowledging a finding and sharing a document that was not requested are not available. Client collaboration sends no notifications and has no concurrency control. Production-like identity proof is pending. | Governance Product Verification record | 2026-11-02 |
Versioned framework and obligation graph CAP-FRAMEWORK-GRAPH Documentation | All plans Live | Releases are authored, reviewed and published by AIGovernr administrators; customers read published releases only. Coverage is mapped published obligations divided by total published obligations: it measures content mapping, not a customer's controls or evidence, and is not certification or legal validation. | Regulatory Content Verification record | 2026-11-02 |
Permission-aware governance copilot CAP-GOV-COPILOT Documentation | All plans Live | Ask is extractive: it returns excerpts from published framework obligations and the findings and risks the user may view, with up to six citations, and abstains when nothing matches. No language model is used for these answers. Drafts are generic templates labelled Draft. The built-in evaluation is a fixed self-check, not a test run; production evaluation and user outcome evidence remain pending. Other AI Assistant tools (gap analysis, remediation suggestions, finding drafts, summaries) are not part of the copilot: they use OpenAI only when the deployment configures a key, are not cited, and gap analysis is unavailable without one. | AI Safety Verification record | 2026-11-02 |
Governed connector ingestion CAP-CONNECTORS Documentation | Not released Planned | Project → Integrations stores encrypted connector credentials, scopes, field mappings and sync history for the catalog providers, but this deployment includes no remote adapters. Connection tests only check that a credential is stored, nothing is pulled from GitHub, Jira, ServiceNow, Slack, Teams or any other provider, sync is disabled in the app, and there is no retry logic. | Platform Integrations Verification record | 2026-11-02 |
OIDC, SAML and SCIM enterprise identity controls CAP-ENTERPRISE-ID Documentation | Not released Planned | There is no federated sign-in through OIDC or SAML; connection details can be recorded and tests only validate the fields. There is no SCIM 2.0 endpoint; provisioning is a project-scoped API that updates existing members. Sign-in domain proof is recorded without a DNS check. Available today on Business, Enterprise and Platinum: access review campaigns, whose revoke decisions remove access. | Security Engineering Verification record | 2026-10-16 Review overdue — do not publish |
Regional processing policy, retention and key lifecycle CAP-REGIONAL-DATA Documentation | Not released Planned | Data Controls stores allowed regions, retention periods, deployment tier, encryption mode and backup settings, and records route checks. Region policy is checked only on connector sync, which no adapter uses; other processing is not region-checked. Key rotation records a version number only; customer-managed keys are not implemented. Run retention soft-deletes expired evidence, agent traces and copilot interactions on demand. Physical regional hosting and private or on-premises deployment are not offered. | Security Engineering Verification record | 2026-10-16 Review overdue — do not publish |
Developer CLI and CI checks CAP-CLI Documentation | Not released Planned | No CLI, CI check or repository scanner is released as a supported product. The evaluation CI workflow template refers to an aigovernr-evaluator command that is not published. | Developer Platform Verification record | 2026-11-02 |
Public developer API CAP-PUBLIC-API Documentation | Not released Planned | The application's endpoints are authenticated by user sign-in and are not a supported public API: there are no general API keys, no published API reference, no versioning and no SLA. Narrow machine endpoints exist only for agent enforcement, monitoring signals and evaluation results. | Developer Platform Verification record | 2026-11-02 |
Runtime agent policy enforcement API CAP-AGENT-ENFORCEMENT Documentation | All plans Live | Your agent or gateway calls POST api/runtime/agents/{projectId}/authorize with an X-Agent-Enforcement-Key created in Agent & MCP Governance, and reports traces to the traces endpoint. Decisions (Allowed, Denied, ApprovalRequired) check the profile's allowed and prohibited actions, explicit per-tool Allow permissions, external authorisation and the spend limit; data, domain and rate limits on the agent card are not checked. Enforcement happens only if the agent calls the API. Calls are refused while the workspace is read-only. No per-key rate limit or SDK is provided. | AI Safety Verification record | 2026-11-02 |
Runtime monitoring signal ingestion CAP-RUNTIME-SIGNALS Documentation | All plans Live | Your system sends POST api/runtime/signals with an X-Monitoring-Key from a Monitoring signal source; each signal must name a governed asset in the project. Signals are idempotent per source and limited to 120 per minute per source (HTTP 429). Threshold breaches open a material change, impact assessment, reassessment case and incident, and can mark linked decisions for review. Source keys cannot be revoked or rotated in the app. No SDK is provided. | Governance Product Verification record | 2026-11-02 |
Repository scanning on GitHub push and pull request CAP-GITHUB-WEBHOOK Documentation | Not released Planned | A signed GitHub App webhook (POST api/github/webhook) records push and pull request events for an already connected project, but events are stored unprocessed: no scan runs and no pull request comment is posted. The repository connection screen is withdrawn until scanning is released. | Developer Platform Verification record | 2026-11-02 |
Shadow AI register, AI tool catalogue and community submissions CAP-SHADOW-AI Documentation | All plans Live | Discovery reads files and CSV exports the team uploads (expense, SSO, dependency files, tool lists) plus the latest compliance scan; there is no live connector to identity providers, expense systems, browsers or networks. Matching relies on the catalogue's signatures, so tools missing from the catalogue are only found by name. Catalogue facts are curated from vendor sources and can go out of date; entries show whether and when they were verified. Risk scores are rule-based and explained factor by factor. Promoting a tool to an AI System counts toward the plan's AI System limit. | Governance Product Verification record | 2026-11-02 |
Evaluation CI workflow template CAP-EVAL-CI Documentation | All plans Live | CI config generates a GitHub Actions template for an approved evaluation specification; AIGovernr does not run evaluations. The template is a starting point, not a working pipeline: it calls an aigovernr-evaluator command that is not published and includes no AIGovernr endpoint or project ID. Results are accepted at POST api/evaluations/results only with an HMAC-SHA256 signature from a per-project key held in AIGovernr server configuration; keys cannot be created in the app. | Developer Platform Verification record | 2026-11-02 |