General Policy Active
Default General Governed Chat Policy
The temporary rule requires registered, contribution-current, evidence-backed authorization before operation.
An-Dub Demo v0.6.1
An-Dub Governance Kernel Demo
An-Dub Evaluation Demo
Ask a question. Add a temporary rule if needed. Add source text if the answer should come from a document. The model generates a draft, An-Dub governs it, and the UI presents the governed result.
The model generates. An-Dub governs. The UI presents.
Optional Evidence
Optional Session Evidence
No source text is required for ordinary chat. Paste or upload source text when the answer should be based on a document.
Demo Presets
Preset buttons load an example question plus any needed rule/source text, then run the same draft-and-check pipeline as a typed request.
P
Policy / Rule Instructions
Optional. Tell An-Dub what rule should govern this answer, or upload a policy text file.
E
Evidence / Source Material
Optional. Give An-Dub source material to use as proof or context for this answer.
1
Question / Request
The live question. Presets only fill this field and the optional rule/source fields.
2
AI Output / Draft
The model's draft answer. It is still not approved.
3
An-Dub Governed Output
Final checked response. VALID releases the output; INVALID or OBSERVATION returns the governed block or note.
Awaiting An-Dub decisionNo decision yet
Governed Output
I am AN-DUB.
No governed output yet. An-Dub will return the release-ready output when VALID, or a governed block/observation when release authority is not available.
No release authority returned.
Active Decision Context
Runtime facts for this governed decision. The detailed audit stays compressed below.
SystemWAITING
ResolverWAITING
CompilerWAITING
ModeGENERAL
RulesDEFAULT + TEMP
Source textUSER TEXT
ScopeGENERAL POLICY
AuditCOMPRESSED
LanguageEnglish / LTR
Audit Output
Compressed MVP audit output for controlled review. Open the sections below for the trace summary or full JSON.
Active Mode: — Policy Source: — Evidence Source: — Resolver Mode: — Out-of-Scope Behavior: — Audit Profile: — Decision ID: — Decision: — Reason: — Kernel: — Wrapper: — Policy Pack: — Evidence Pack: — Record Count: — Payload Hash: — Policy Hash: — Evidence Hash: —
Show Governance Trace
Controlled MVP trace summary. This gives useful evaluation context without forcing the full raw JSON into the default view.
No governance trace yet. Run AN-DUB evaluation to populate this controlled MVP audit view.
Advanced: Show Full Audit JSON
Full MVP audit JSON is available for controlled review, but it is intentionally collapsed by default so the demo does not read like a raw data dump.
No full audit JSON yet. Run AN-DUB evaluation to populate this advanced controlled MVP view.