AI assurance assessment.
Review the scope, status and evidence behind Honua’s AI security assessment.
Assessment recorded: 2026-10-25 · Review due:
Assessment in progress. No installed release is qualified by this assessment. This dated snapshot must be reviewed after the due date.
Assessment scope
Honua Server; Studio / Console; SDK agent integrations; DevOps agent; Deployment infrastructure; Development-agent workflows. We assess agent permissions and access boundaries within supported deployments.
Excluded: Cross-tenant isolation; External certification; Unconfigured model providers.
Customer-managed deployments require operator evidence for identity configuration, provider terms, network restrictions, secrets, retention and recovery. Any Honua-managed deployment requires its own operational evidence.
Framework alignment status
Honua is assessing its AI capabilities against the frameworks below. These mappings identify work to verify; they do not establish framework equivalence, full compliance, or certification. Honua does not claim ISO/IEC 42001 certification or FedRAMP authorization.
| Framework | Version | Status | Coverage |
|---|---|---|---|
| NIST AI RMF | 1.0 (2023) | Assessment in progress | 4 assessment topics mapped; verification pending |
| NIST Generative AI Profile | AI 600-1 (July 2024) | Assessment in progress | 12 assessment topics mapped; verification pending |
| NIST agent identity concept paper | Draft, February 2026 | Assessment in progress | 5 assessment topics mapped; verification pending |
| OWASP Top 10 for Agentic Applications | 2026 (December 9, 2025) | Assessment in progress | 10 assessment topics mapped; verification pending |
| ISO/IEC 42001 | 2023 | Assessment in progress | 1 assessment topic mapped; verification pending |
| NIST SP 800-53 | Rev. 5, catalog release 5.2.0 | Assessment in progress | 20 assessment topics mapped; verification pending |
The NIST agent identity concept paper is a draft with five focus areas: identification, authorization, delegation, logging and data flows. It is distinct from the broader AI Agent Standards Initiative. ISO clause-level assessment awaits authorized standard text; SP 800-53 individual controls and enhancements require tailoring.
Thematic draft mapping of functions, risk categories, focus areas and control families. No framework equivalence or clause-level conformance is asserted.
Control assessment
| Control | Area | Verification | Evidence |
|---|---|---|---|
| AI-01 | Governance and accountability | Pending | Evidence collection pending |
| AI-02 | AI inventory and data flows | Pending | Evidence collection pending |
| AI-03 | Provider transport and disclosure | Pending | Source: server |
| AI-04 | Agent identity and delegation | Pending | Source: mcp; Source: console |
| AI-05 | Protected actions and human approval | Pending | Source: approval |
| AI-06 | Untrusted context and output | Pending | Source: studio; Source: sdk |
| AI-07 | Worker containment and supply chain | Pending | Evidence collection pending |
| AI-08 | Budgets and bounded execution | Pending | Source: server; Source: sdk |
| AI-09 | Audit and incident reconstruction | Pending | Source: console |
| AI-10 | Quality and impact evaluation | Pending | Evidence collection pending |
| AI-11 | Incident response and recovery | Pending | Evidence collection pending |
| AI-12 | Qualification and truthful claims | Pending | Evidence collection pending |
Public evidence
The following links pin inspected source versions. They demonstrate inspectable implementation locations, not successful model evaluations or installed deployment behavior. Runtime qualification, model evaluations, operational exercises and independent review remain pending. No qualification receipts are available from this assessment yet.
- honua-server: StudioAiProxyConfiguration.cs — 2026-10-25. Source reference only; not installed execution evidence.
- honua-server: McpAuthorizationHelper.cs — 2026-10-25. Source reference only; not installed execution evidence.
- honua-server: ProposalEndpoints.cs — 2026-10-25. Source reference only; not installed execution evidence.
- honua-studio: orchestrator.ts — 2026-10-25. Source reference only; not installed execution evidence.
- honua-sdk-js: session.ts — 2026-10-25. Source reference only; not installed execution evidence.
- honua-console: AgentSessionBff.cs — 2026-10-25. Source reference only; not installed execution evidence.
Download the assessment mapping (JSON). The download includes framework topics, control mappings, dates and the same public source references.
What remains to verify
Provider disclosure policies, delegated authority, protected actions, untrusted context, execution limits, audit records, supply-chain integrity, recovery and geospatial correctness require scoped verification. Management-system approval, training and independent assessment evidence are also pending.
Enabling external models can send prompts, supplied context, report summaries or imagery to the configured provider. Operators must approve the destination and data handling for each enabled use case.
Material changes to models, tools, policies or deployment configuration require reassessment. Missing, skipped, stale and inconclusive results do not count as passes.
Request restricted assessment material through security@honua.io. Availability depends on completion and disclosure review.