Independent policy kernel for AI agents: judge before execute. F1–F13 floors, VAULT999 receipts. Not a model. Not a chatbot.
FAQ (answers come from our records)
How do I install it?
The repository provides this install command: pip install arifos. It comes from the repository itself; we cannot guarantee it runs as-is in your client.
Who is it for?
We filed it under these tags: 开源项目. The tags come from the entry itself, not a taxonomy we imposed.
Who maintains it?
repository ariffazil/arifOS;first catalogued 2026-10-10;last seen 2026-10-10。There is no official certification or quality rating here — we record sources and dates only.
> ### ⚠️ Current verified state (2026-10-11) — measured, not intended
以下摘录自该仓库的 README(片段) · 在 GitHub 读完整 README →
>
> Re-probed live by FI-008 on this date. This block states only what was observed; it does not replace the hand-audited manifest below.
>
> **Two properties hold end-to-end:** deterministic rejection (the judge refuses without consulting a model — gate: hard_deterministic, llm_consulted: false) and prevention of an unauthorised mutation.
>
> **Four do not:** verified actor (the signing path exists and the lane on :18900 listens, but observed calls arrived **unsigned**; actor_verified:false) · receipt fidelity **on the invocation-telemetry surface** (/var/lib/arifos/metrics/tool_invocations.jsonl, 194,243 lines, **0 governance fields**; a call whose kernel verdict was HOLD was persisted as ok:true) · sink agreement (one event: ledger ok:false, organ counters omit it) · independent reconstruction (no external verifier bound).
>
> **Note on two witness surfaces:** the VAULT999 verdict ledger (outcomes.jsonl, 95,091 lines, 89,509 carrying a verdict/decision field) is a **different object** from the invocation-telemetry ledger above. They share a shape and not a schema. A verdict is not transferable between them.
>
> **Floors: pass 2 · fail F2/F11/F12/F13 · unmeasured 10. Unmeasured is not pass.**
>
> No customers, no pilots, no signed contracts, no independent benchmark, no external security audit. Current revenue value: USD 0.
>
> Claim-level audit: AAA/forge_work/grant-arifos-reality-reforge-2026-10-11/.
= 0.0" — every finite score passed); that earlier pass count was a FALSE PASS, not a loss. UNMEASURED IS NOT PASS.
federation_schema: 2.0.0
mcp_protocol: advertises 2026-07-28; live initialize and the internal conformance runner settle on 2025-11-25 (supported: 2026-07-28 · 2025-11-25 · 2025-03-26 · 2024-11-05)
organs: 7 per the ratified organ table (FEDERATION_CONTRACT §2) + plane classes for boundary services (see Architecture)
vault999: healthy (457K+ records, append-only)
contract_status: 8/8 published schemas, contract_drift: false
tool_manifest_url: https://arifos.arif-fazil.com/tools.json (37,046 bytes, live)
apex_zen: A2A delegates ⊥ MCP equips ⊥ ACT mutates ⊥ arifOS governs ⊥ F13 decides
truth_rule: live :8088/health + tools/list beat any static count in prose — re-run before you quote this file
generated_by: scripts/update_readme_sot.py re-stamps commits, drift, last_verified and the vault count only; every other field here is verified by hand at audit time and must be re-verified the same way
holds: Merkle signing lane · WELL biometrics · medical purge (Pilihan A)
seal_readiness_gaps: graphiti=RETIRED_888(2026-09-04) · semantic_floor=off_by_choice(ARIFOS_ML_FLOORS=0) · langfuse=SOVEREIGN_CUTOVER→kabarkan(live)
-->
arifOS — The Authority Plane of the arifOS Federation
**arifOS evaluates consequential AI actions against constitutional floors and returns an independent verdict _before_ execution occurs.**
When an AI agent proposes to write, delete, deploy, or spend, arifOS inserts a constitutional judgment step: the agent proposes, arifOS evaluates the proposal against the F1–F13 floors, a verdict is returned, and only then does execution proceed. Every consequential verdict is written into an append-only hash chain with its decision context, evidence references, provenance and receipt hashes.
In a world where intelligence is abundant, authority becomes the scarce resource. arifOS exists to keep judgment independent from execution.
**This is not an AI model. It is not an agent framework. It is a constitutional authority system — the layer between "agent wants to act" and "action is permitted."**
Two invariants hold everywhere in this document:
> **Proposer ≠ Judge ≠ Executor ≠ Witness**
> **Human sovereignty > machine authority**
**What arifOS is not:** not an AI model · not an agent framework · not an execution engine (A-FORGE executes) · not an attention plane (AAA attends) · not the independent witness (FRAME witnesses; VAULT999 remembers) · not a substitute for authentication, sandboxing, or legal review.
| Audience | What you get |
|---|---|
| **Human** | A quiet veto: the agent proposes, the kernel records a verdict, you stay sovereign |
| **Agent / A2A** | MCP tools + receipts. You do not get the keys. A2A v1.0 (a2a_version: 1.0.1 in the agent card), not v1.2 — discovery is owned by AAA, judgment by arifOS, **execution by A-FORGE** |
| **Institution** | Policy floors F1–F13, a hash-chained VAULT999 audit trail, model-vendor independence |
| **Indexer / Search** | Structured metadata, llms.txt, tools.json, CITATION.cff |