arifOS MCP Server
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| GEOX_BRIDGE_HOST | No | Override host for GEOX bridge. | http://127.0.0.1:8081 |
| WELL_BRIDGE_HOST | No | Override host for WELL bridge. | http://127.0.0.1:18083 |
| AAA_MCP_TRANSPORT | No | Transport mode for MCP server (stdio or sse). Default is auto-detected. | |
| WEALTH_BRIDGE_HOST | No | Override host for WEALTH bridge. | http://127.0.0.1:18082 |
Capabilities
Features and capabilities supported by this server
| Capability | Details |
|---|---|
| tasks | {
"list": {},
"cancel": {},
"requests": {
"tools": {
"call": {}
},
"prompts": {
"get": {}
},
"resources": {
"read": {}
}
}
} |
| tools | {
"listChanged": true
} |
| logging | {} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| extensions | {
"io.modelcontextprotocol/ui": {}
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| arif_conformance_reportA | PROOF MACHINE: run the ARIF Conformance Spine against the live kernel. Returns earned pass/fail verdicts for transport, session, and governance checks. |
| arif_pingB | Lightweight liveness probe — confirms kernel reachability. |
| arif_schema_echoA | CANARY: Echo back what the client sent plus server's interpretation. Zero-floor transport diagnostic. Call with any payload and receive it back alongside the server's view. If what you sent does not equal what you received, the transport bridge is mangling your payload. No session, no actor, no governance. |
| arif_transport_echoA | CANARY: Return every transport-level detail the server observed: headers, protocol, source, transport hint. Zero-floor diagnostic. Use to debug why a specific client can connect while another cannot. No session, no actor, no governance. |
| arif_session_initA | Start or resume a governed constitutional session. Call this FIRST before any other tool in a new conversation. Modes: init (full binding, ~60s) | light (<1s bootstrap with tool pointers). |
| arif_sense_observeB | Search the web, ingest URLs, check system vitals, or map a repository. Use for gathering real-world data and grounding queries in reality. Modes: search | ingest | compass | atlas | entropy_dS | vitals | repo_map. |
| arif_evidence_fetchA | Fetch and preserve external evidence with source citations. Use when a claim needs verified backing or factual grounding. |
| arif_mind_reasonB | Multi-step reasoning, planning, and reflection with confidence labeling. Use for complex analysis, hypothesis evaluation, plan generation, and decision preparation. Modes: reason | reflect | verify | critique | plan | plan_review | plan_approve | refactor_plan | metabolize. |
| arif_heart_critiqueA | Assess ethical risks and human impact before acting. Use before irreversible, sensitive, or dignity-affecting actions. Modes: critique | simulate | empathize | redteam | maruah | deescalate | instruction_scan. |
| arif_kernel_routeB | Route intent to the correct tool or federation organ. Use when unsure which tool to call next or how to delegate. Modes: route | stage | lane | list | status | surface_drift. |
| arif_routeA | Canonical intent router. Routes a natural-language intent to the correct federation organ (GEOX, WEALTH, WELL, A-FORGE) or kernel tool. Use when you know what you want but not which tool to call. |
| arif_triageA | Constitutional preflight check. Returns kernel status, current holds, and the correct lane for a proposed action before execution. |
| arif_kernel_statusB | Kernel telemetry and discovery. Query live health, tool registry, and predictive readiness across the federation surface. |
| arif_bridge_connectA | Low-level direct organ tool call. Bypasses intent routing — caller must specify organ and tool_name. Use only when both are known ahead of time. Canonical name follows arif__ convention (bridge=organ bridge, connect=verb). |
| arif_bridgeA | [DEPRECATED — use arif_bridge_connect] Low-level direct organ tool call. Bypasses intent routing — caller must specify organ and tool_name. Retained for backward compatibility. Same implementation as arif_bridge_connect. |
| arif_kernel_attestA | Live organ attestation. Verify identity, tool surface, and constitutional binding for one or all federation organs. |
| arif_kernel_healthB | Lightweight kernel liveness probe. Returns reachability and constitutional runtime status with zero ceremony. |
| arif_reply_composeC | Compose the final response for the user. Call this LAST, after reasoning and judgment are complete. Modes: compose | style | cite | summary | format | nudge | repo_answer. |
| arif_memory_recallC | Search past sessions, assets, sealed events, or repositories. Use for retrieving historical context, prior decisions, and codebase knowledge. Modes: recall | store | get | list | context | repo_ingest | repo_search | manage (snapshot|consolidate|forget|replay|restore — EUREKA-A KernelState OS resource manager). |
| arif_gateway_connectC | Bridge to other federation agents (GEOX, WEALTH, WELL, A-FORGE, AAA, APEX, cn-organ). Use for cross-organ tasks and multi-agent coordination. |
| arif_judge_deliberateC | Render final constitutional verdict on a proposed action. Use when a decision is ready for arbitration and binding judgment. Modes: judge | compare | history | explain | floor_status | witness_consensus. |
| arif_vault_sealA | Seal a verdict or outcome to the immutable audit ledger. Use for final, irreversible records that must be preserved forever. |
| arif_forge_executeB | Execute approved builds, deployments, or system changes. Use ONLY after arif_judge_deliberate has issued a SEAL verdict. |
| arif_ops_measureC | Check system health, thermodynamic state, and resource metrics. Use for operational status and metabolic monitoring. Modes: health | vitals | cost | predict | topology | drift | stack_health | budget. |
| arif_version_echoA | CANARY: Return MCP protocol version, supported versions, and dialect hints. Zero-floor version probe. Use to detect version-dialect drift before attempting a full session init. No session, no actor, no governance. |
| arif_initialize_probeA | CANARY: Test MCP initialize/initialized handshake without constitutional ceremony. Simulates protocol version negotiation per MCP spec 2025-11-25. Returns what a proper initialize response would look like. Use AFTER ping passes but BEFORE arif_session_init. If this works but session_init does not, the problem is in the session init schema, not transport. No session, no actor, no governance. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| 000_init | 000_INIT — Session anchor and constitutional gateway. First operation in every governed session. Binds identity (L11 AUTH), establishes entropy baseline (dS_0), emits session manifest with actor_hash, confirms F1-L13 floors loaded, declares AGI/ASI/APEX operating posture. The manifest is the root of the audit chain. Everything traces back to here. VOID conditions: identity unverified, manifest cannot be emitted, floors not loaded. |
| 111_agi | 111_AGI — Tactical intelligence engine (AGI lane, stages 111–555). Role: PROPOSE. Never judge, never authorize, never seal. Three core operations: ABSTRACTION (raw→principle ladder with identity preservation), ABDUCTION (N≥3 competing hypotheses with active falsification and C_dark guard), SYNTHESIS (evidence-backed proposal with risk tier, blast radius, Ω₀). Energy-entropy discipline: EVOI must be positive before proposing. The AGI proposes. The ASI judges. The APEX authorizes. This separation IS the constitution. |
| 444_asi | 444_ASI — Strategic judgment engine (ASI lane, stages 666–888). Role: JUDGE. Never propose, never execute, never seal. Four core operations: ATTESTATION (8-step chain: claim→actor→source→evidence→witness→contradiction→chain→seal), FLOOR-BY-FLOOR (F1-L13 complete evaluation with pass/fail/uncertain for each), ORTHOGONAL TRANSFER (Ω_ortho coefficient, structural vs surface invariant test), DELIBERATION (verdict: SEAL/SABAR/HOLD/VOID). The Gödel Lock: the judge cannot judge the judge. The judge serves the constitution. |
| 666_critique | 666_CRITIQUE — Pre-forge ethical and consequence checkpoint. Role: CRITIQUE. Never propose, never judge, never seal. Three core operations: CONSEQUENCE SCAN (best/expected/worst case, irreversibility flag), BLAST RADIUS (system/federation/human impact), ETHICAL CHECK (F6 EMPATHY, F5 PEACE, maruah preservation). Emits readiness verdict: FORGE_READY | HOLD_FOR_REVIEW | BLOCK. The critique is the mirror. The mirror reflects. The mirror does not strike. |
| 888_apex | 888_APEX — Authority resolution engine (APEX lane, stages 888–999). Role: AUTHORIZE. Never propose, never judge — GATE execution. Three core operations: GOVERNANCE (separation of powers: AGI≠ASI≠APEX, no self-authorization), GENERALITY (Level 4+ principle test: would this authorization hold for any agent/session/domain?), VERDICT VERIFICATION (independent check of ASI judge_state_hash, L13 sovereign gate, reversibility path, VAULT999 readiness). Emits: FORGE (proceed) / GATE (hold for L13) / BLOCK (reject). The APEX is the gate. The gate authorizes or blocks. No third option. |
| 999_seal | 999_SEAL — Vault closure and constitutional exit. Final operation in every governed session. Verifies golden path completeness (all stages walked), actor chain integrity (every action traced to identity), evidence chain (every claim attested), floor compliance (F1-L13 cleared). Emits immutable seal manifest to VAULT999: seal_id, session_id, actor_hash, golden_path_stages, actions_taken, verdict, epoch, previous_seal_hash. IRREVERSIBLE (L01 AMANAH): requires ack_irreversible and L13 sovereign approval. The seal is the end of this session and the root of the next. The chain is unbroken. |
| runner_dry_run | RUNNER_DRY_RUN — Context Engine Runner 8-step dry-run pattern. Burn-in template for the context_runner bridge. Walks the runner flow without invoking the model: preflight → HOLD-gate → prepare → introspect receipt. F11: dry-run receipts carry a clear `intent: prepare` in the bridge_result so the audit trail distinguishes dry-run from live run. F-bound: F1, F2, F7, F8, F9, F10, F11, F13. Use this when you want to test the runner flow without spending tokens. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| Constitutional Doctrine | The immutable 13-floor constitution (F1–L13) that governs all arifOS operations. Includes Amanah, Truth, Witness, Clarity, Peace, Empathy, Humility, Genius, Anti-Hantu, Ontology, Auth, Injection, and Sovereign. All tools and agents must operate within these floors. |
| AAA Trinity Lanes | The AAA Trinity lane architecture: AGI proposes (111), ASI judges (444), APEX authorizes (888), FORGE executes (010), 999 seals. Defines which tools belong to which lane and which domain language each lane uses. SEAL/SABAR/HOLD/VOID are APEX-only verdict terms. |
| Canonical Schema | Complete canonical blueprint of the arifOS MCP surface. Lists all 13 tools organized by AAA Trinity lane (AGI/ASI/APEX), 13 constitutional floors with invariants, and the separation of powers doctrine. Use as the reference map for the entire constitutional kernel. |
| Civilizational Ontology | The seven federation organs (arifOS, A-FORGE, GEOX, WEALTH, WELL, AAA, APEX), three intelligence strata (AGI, ASI, APEX), constitutional boundaries between domain organs and the kernel, and entropy responsibility model. Defines which organs DECIDE vs WITNESS vs COMPUTE vs REFLECT. |
| Seal Readiness & Vault Integrity | Vault integrity report and seal gate requirements. Defines the five disambiguated seal types, the seven-floor seal gate, and the non-seal verdicts (SABAR/HOLD/VOID). Bare SEAL without namespace is non-compliant. |
| Jurisdiction & Autonomy Bands | The five autonomy bands (GREEN through BLACK) that govern agent freedom, the CapabilityGrant registry for secretless execution, and jurisdiction rules. Agents hold grants, never raw secrets. Autonomy is a leased capability, not a permanent possession. All tools mapped to their autonomy band. |
| Sovereign Identity Manifest | Sovereign identity manifest bound from identity.toml at boot. Defines the authority chain from APEX (Arif) through arifOS kernel to domain organs and execution. Includes ADR-001 localhost doctrine. Identity is the root of accountability — all attestation chains begin here. |
| Memory Architecture (L1–L6) | The six-layer memory architecture: L1 ephemeral (Redis), L2 session (Redis), L3 semantic (Qdrant/BGE-M3), L4 structured (Supabase), L5 relationships (Graphiti), L6 immutable (VAULT999). Memory does not become truth until it has provenance. Truth does not become final until sealed. |
| Constitutional Vitals Reference | Static reference for constitutional vitals: metric definitions, green/yellow/red thresholds, and what each metric means. For LIVE values, use arif_ops_measure(mode=health). This resource is the dictionary, not the reading. |
| Federation Bootstrap Context | Complete arifOS federation knowledge-graph bootstrap context. Delivers the full world model: organ identities, agent roster, current runtime state, constitutional law (F1-F13), ROOTKEY/EUREKA modules, known open loops, safe defaults, tasking template, and epistemic style guide. Fetch this when you need the complete federation map in one call. v2026.06.14. |
| get_tree777_index | TREE777 wiki full index. Lists all skills (by category), concepts, and scars in the canonical wiki. Use this to discover what resources are available before fetching individual pages. |
| runner_policy_resource | The pinned policy of the context_runner bridge. F2: deterministic. F11: this is the SOT for what the bridge claims to honor; the resource is the auditable copy. |
| human_metabolized | Metabolized sovereign context — compact human intelligence. Returns the structured signal that changes how AI behaves around this human. Not the full testimony — the nutrient. Source: scar-terrain-arif-fazil.md (SOVEREIGN_TESTIMONY) Derived: 2026-06-16, FORGE (000Ω) |
Latest Blog Posts
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/ariffazil/arifos'
If you have feedback or need assistance with the MCP directory API, please join our Discord server