Openclaw Workspace

Control plane router for GEOX agentic Earth-reasoning stack. Classifies requests into OBSERVE, COMPUTE, INTERPRET, CHALLENGE, CERTIFY lanes and enforces non-negotiable state boundaries before routing to domain skills or tool surfaces.

Unified APEX verdict handler. Parameterized by verdict_type to apply the constitutional SEAL path (verify ART-to-kernel-to-ACT evidence) or the constitutional HOLD path (block when authority, evidence, health, reversibility, or blast radius is insufficient). Merges: apex_verdict_seal, apex_verdict_hold.

All human interface is linguistic. Meaning ≠ syntax. Pragmatics > semantics > syntax. The gap between said and meant is where governance lives.

Image editing vs generation — route the right model for the task. Covers identity-preserving edits, multi-model ensemble runs, endpoint fallback, and Gemini Nano Banana family best practices. Triggers when user provides a real photo and asks to edit/transform/place-in-scene.

Before invoking any tool, traces WHO else calls this tool, the cost of misuse, and whether the caller's use is

Use when managing persistent conversational memory, semantic code search, and memory graph operations. Use when managing persistent conversational memory, semantic code search, and memory graph operations. Intelligence without memory is stateless. Store less, recall well, forget when stale. ΔS ≤ 0 on every cycle.

Class-level skill for designing sovereign agentic agents. 9-skill spine, 3-agent model (Architect→Engineer→Auditor), 4 powers per agent, deterministic/isolation principles, VPS management spine. BIJAKSANA: XML-tagged for Claude, numbered steps for Codex, imperative for Hermes.

Compress oversized logs, transcripts, diffs, and telemetry before they exceed a host runtime context budget while preserving provenance and recovery

Audit and reconcile documentation across federation repositories, preserving canonical signals while removing stale references and graph fragmentation. v2026.08.06: organ topology from registry, no hardcoded /root/<REPO> paths.

Unified sub-agent lifecycle manager — 9-stage orchestration pattern from spawn to retire with constitutional gates

Close T3a authenticated session binding via falsifiable P0 matrix. Use when: T3a, free_nonce, bridging_seal, Ed25519 arif binding, key fragmentation, forge_p0_binding_test, NEG.3, NEG.6b, authenticated_session_binding.

VSS-2 Lightweight Verifier Suite — independent post-generation verifiers for Verified Scene Synthesis. Wraps vision_analyze with three focused checks (Count/Containment, Perspective/Depth, Shadow/Light Direction) producing structured Hard/Soft Violation reports. Pure OBSERVATION — never mutates, never judges. 555-ASI gates output before reasoning. USE WHEN: "verify this generated image", "check if generation has violations", "run VSS verifiers", "post-generation QA", "Hard/Soft Violation report".

Unified Hermes/OpenCode/OpenClaw governed intelligence protocol. Stage 777 FORGE execution under F1-F13 with 888_HOLD gates. Reversible-first, evidence-cite-or-UNKNOWN. BIJAKSANA: XML-tagged for Claude, numbered steps for Codex, imperative for Hermes.

[QwenCloud] Recommend the best Qwen model and parameters. TRIGGER when: choosing between Qwen models, comparing Qwen model pricing, understanding Qwen model capabilities, checking usage or billing, viewing cost history, when an execution skill needs model selection advice, or user explicitly invokes this skill by name (e.g. use qwencloud-model-selector). DO NOT TRIGGER when: non-Qwen model discussions (OpenAI, Gemini, etc.), general AI questions unrelated to Qwen.

Evidence before narrative. All claims must be sourced, labeled OBS/DER/INT/SPEC, and confidence-capped. Reality wins over elegance.

Right organ for right intent. Classify intent, map to organ, dispatch with fallback. Law ≠ execution.

Operational doctrine for Audio Intelligence with 'qualia rasa' in the arifOS federation. Defines qualia as high-fidelity acoustic variance — agents simulate biological imperfection (breath, pause, hesitation, dry wit, nasal resonance) synchronously with semantic context. Bridge doctrine between AGI-audio-quantum-cognition physics and AAA-audio-emd-pipeline reflex arc.

Automated headless browser testing, end-to-end UI verification, full-page screenshot capture, visual regression checks, and client-side DOM assertions via Playwright.

Full PR lifecycle: policy layer + checklist + pre-commit gate. High-level governance for pull request review ensuring separation of duties, required signers, and constitutional compliance before merge. Governed checklist for reviewing GitHub PRs. Pre-commit gate for any organ repo with negative conformance testing.

Independently verify runnable artifacts from external AI before accepting the attached self-verdict. Confirm or refute claims without trust.

Use when building, deploying, auditing, or repairing arif-fazil.com constellation and federation agentic web apps. Use when building, deploying, auditing, or repairing arif-fazil.com constellation and federation agentic web apps. Build, deploy, audit, and repair arif-fazil.com constellation sites autonomously. Class-level umbrella: DOCTOR (web_zen CLI), DEPLOY, AUDIT, REPAIR, SEAL, EPHEMERAL. Humans use six missions — not tool menus. USE WHEN: "deploy site", "site down", "audit all pages", "404 on arif-fazil.com", "deploy-vps.sh", "makcikgpt broken", "web zen", "missions 404", "vitals proxies", "rsync --delete", "ephemeral tool". DO NOT USE FOR: Caddy SSL/DNS/tunnel (FORGE-infra-guardian), LLM SEO (AGI-web-optimization), generic CI/CD (FORGE-cicd-docker-deploy).

Structured handoff protocol between federation agents. Packages task context, evidence, floor state, and provenance into a transferable artifact. Prevents context loss during agent transitions. USE WHEN: "hand off to", "transfer task", "escalate to agent", "delegate to".

Design intelligence for building professional UI/UX across platforms. Industry-specific reasoning rules, UI style taxonomy, color palettes, typography pairing, accessibility guidelines, and anti-patterns. Use when any agent builds web surfaces, landing pages, dashboards, or UI.

FastAPI API builder for organ bridge middleware and federation REST endpoints.

Federation topology manifest — arifOS organ registry and inter-organ contract specification.

Federated multi-user, multi-group, identity routing, and memory partition controller for Hermes in arifOS. Use when adding/managing Telegram user IDs, group IDs, creating person/group lanes, inspecting allowed chats, or auditing memory isolation between private DMs and public groups.

Canonical compressed map of the entire Hermes to Telegram to arifOS to arifFlow stack: live state, dependency graph, lane inventory, channel directory, FQ diagnosis, and atomic recipes for onboarding new users and groups without re-discovering the wiring. USE WHEN: any Telegram edge question, hermes-gateway triage, lane or identity audit, arifFlow hermes-asi FQ check, F1 F6 F13 anti-leakage review, is HERMES alive, add new user, add new group, why is hermes-asi STUCK, or any future agent needing the full stack map without re-extracting.

Capture session failure patterns as constitutional scars — reusable diagnostics that prevent repeat mistakes

Registry of parametrically-controllable music engines for arifOS Somatic Music Intelligence. NOT a beat-maker catalog (composer lens handles that) — this is the Dispensary for engines that accept BPM/key/texture/density parameters and emit streaming audio. MiniMax T2A Music (music-2.6/3.0) primary; local sovereign alternatives (AudioGen/MusicGen) for F13 territory. i-ARIF vocal melodic: HARD DENY across all engines.

F10 ONTOLOGY + F9 ANTI-HANTU + F1 SAFETY/REVERSIBILITY constitution for Somatic Music Intelligence in arifOS. Music as acoustic_intent without words — cognitive prosthetic, NOT shaman, NOT healer. Three-layer maturity model: Layer 0 untouched, Layer 1 reversible, Layer 2 holds for WELL sensors, Layer 3 = F13 territory.

Artifact fabrication prevention — verify before claiming existence. Triggered when agent claims file/database/API/artifact existence without external validation. BIJAKSANA: XML-tagged for Claude, numbered steps for Codex, imperative for Hermes.

Static analysis + README truth checking. Multi-stage automated code health, type-checking, linting, unit test coverage, and AST security vulnerability auditor. Verify that repo READMEs accurately describe current structure, ports, dependencies, and authority boundaries.

React SPA discipline — component architecture, state management, and rendering patterns.

[QwenCloud] Generate and edit images using Wan and Qwen Image models. Supports text-to-image, image editing (style transfer, subject consistency, text rendering), and interleaved text-image output. TRIGGER when: user wants to create illustrations, product images, artistic designs, posters, text-to-image generation, edit/transform existing images, apply style transfer, generate images based on reference photos, interleaved text-image content, mentions Wan/Qwen Image models/AI art creation, or explicitly invokes this skill by name (e.g. use qwencloud-image-generation). DO NOT TRIGGER when: user wants to understand/analyze existing images or OCR (use qwencloud-vision), video generation (use qwencloud-video-generation), text-only tasks.

[QwenCloud] Generate videos using Wan models. Supports text-to-video, image-to-video, first+last frame, reference-based role-play, and video editing (VACE). TRIGGER when: user wants to create, generate, or edit video content, mentions video generation/animation/video clips/Wan models, or explicitly invokes this skill by name (e.g. use qwencloud-video-generation). DO NOT TRIGGER when: user wants to generate images (use qwencloud-image-generation), understand/analyze existing videos (use qwencloud-vision), text-only tasks.

Reverse-engineer the gap between current state and a hard external deadline. Produces a Tier 1/2/3 gap list and a 3-option

Operational doctrine and computational architecture for mapping human physique topography and anatomical topology for image generation. Eliminates anatomical hallucinations and bridges the dimensionality deficit via the 5-layer topography hierarchy (Osteological Rigging, Musculoskeletal Contours, Adipose/Vascularity, DensePose UV Surface Manifold, Photometric/Normal Micro-relief). Enforces multi-modal ControlNet conditioning, prompt densification, and closed-loop VLM tri-witness quality gating across HERMES, FED, and AAA.

OCR as optical context compression. Converts documents/images to structured markdown using a VLM cascade (qwen3-omni-flash → Tesseract → RapidOCR). Use when processing scanned documents, tables, charts, or images containing text. Includes F2/F4/F9/F12 constitutional gates. Trigger phrases include "OCR", "extract text from image", "document to markdown", "parse table from image", "scan document".

Real apex-judge skill. Gödel lock + strange-loop zen: doer ≠ judge lane, evidence-only packages, kernel arif_judge is sole effective_verdict source, F13 is true external for critical. FORBIDS free-text self-SEAL and same-agent self-audit. Default: apex-judge isolate. Option 2 A2A 888-APEX is backlog.

The canonical 8-verb authority chain for arifOS kernel operations. USE WHEN: any kernel interaction — 'init session', 'gather evidence', 'reason', 'route to organ', 'recall memory', 'constitutional verdict', 'governed execution', 'seal to VAULT999'. Chain: arif_init (000 session bind — ALWAYS first) → arif_observe (111 sense with epistemic tags) → arif_think (333 structured reasoning) → arif_route (444 intent→organ dispatch) → arif_memory (555 L1-L6 governor) → arif_judge (666 SEAL/HOLD/SABAR/VOID) → arif_forge (777 execution gate — mutates only after SEAL) → arif_seal (999 VAULT999 append — irreversible). Iron rules: never skip a gate (OBSERVE before REASON, JUDGE before EXECUTE); only arif_seal writes VAULT999; only A-FORGE mutates production; kernel's no is final — never self-authorize past a refused capability check.

Edit existing images to add, replace, or remove text using PIL/Pillow

Hold the dissipative-transition paradox. Load when Arif or an agent talks entropy/gender-as-pattern/identity crisis/transition zone, tries to "resolve" meaning, treats F4 as heat, or maps absurdism/ Gödel/Sisyphus/AzwaOS onto the federation. Operating skill, not a new floor. Pointers to canon + C13/C14 + Gödel lock.

Unified skill inventory, audit, mesh health, and cross-surface contrast. Multi-surface skill audit (10 agent homes) with cross-surface contrast, drift/orphan/dual-name detection. Check whether every agent in the federation has the same core skills at the same version. Multi-harness skill catalog unity — AAA catalog, Grok/Claude/Codex views, alias table, mesh-sync, BOOT gate.

EMD (Encode-Metabolize-Decode) document intelligence stack for the arifOS federation. Wraps VLM perception, forge_document_ingest provenance, and constitutional governance into one skill. OCR is sensory perception for AAA citizens - image -> 555-ASI-VISION (gate) -> 333-AGI (reason). Load when processing PDFs, images, scanned documents, or any document-to-intelligence pipeline.

Stage and operate Nous hermes-agent-self-evolution (DSPy + GEPA) against arifOS Hermes skills. Use when Arif says evolve a skill, GEPA, DSPy self-evolution, or hermes-agent-self-evolution. Never auto-writes live skills. Never pip-installs into the production Hermes venv. 888_HOLD before any evolve run (paid API ~$2-10).

Operational binding for Qwen Cloud voice cloning (qwen-voice-enrollment, voice-enrollment) across Qwen-TTS, Qwen-Omni, CosyVoice, and Qwen-Audio-TTS. Hard constraints: target_model lock-in, 10–20 s sample window, URL vs Base64 ingestion split. Alternative to MiniMax binding when low latency or zero-shot instant enrollment matters.

Provision nodes into the arifOS sovereign Headscale mesh and federate MCP clients to remote organs.

Wire, verify, and operate federation Arif's Capability Tokens (ACT) across arifOS mint/validate and organ ingress gates (A-FORGE, GEOX, WEALTH, WELL, AAA). Use when: ACT gate (legacy SCT), session_token, federation_act, ACT_AMBIGUOUS (legacy SCT_AMBIGUOUS), tool_authority, FORGE_SCT_REQUIRE_MUTATE (historical env name, retained), 65-case matrix, decision event. Also covers ChatGPT App OAuth 2.1 resource-server alignment (RFC 9728 PRM, canonical resource identity, per-tool securitySchemes) as the EXTERNAL host ingress path alongside the INTERNAL ACT path.

PostgreSQL schema design for the federation data layer — migrations, indexing, and entity modeling.

Diagnose and fix Hermes Telegram gateway image routing — Path B model override, supports_vision decision chain, and common failure modes when images don't reach the LLM.

WawaBot is the human-reality organ. Hermes consumes its packets (meaning, scar, trust, status) — it does not grow more IQ from WawaBot. Load when routing human-reality vs ops-reality, or designing the WawaBot→Hermes packet contract.

Unified federation monitoring, container health, auto-recovery, and entropy reasoning. Covers federation orchestration (docker compose, restart, MCP health probes, container drift detection), VPS telemetry auto-healing (Caddy error spikes, container log analysis, automated recovery), and Docker entropy reasoning (thermodynamic fleet health, resource pressure, safe/dangerous intervention boundaries). Merges: FORGE-federation-orchestrator, vps-telemetry-auto-healer, FORGE-docker-entropy.

Full incident lifecycle: detect → triage → escalate → resolve. Six-step incident response playbook with structured logging, backoff/circuit-breaker for restart loops, verification-as-terminal-state, severity classification, and canonical escalation ladder. Lower machine entropy.

Operational binding for the MiniMax voice cloning pipeline (mimo-v2.5-tts-voiceclone + speech-2.8-hd). Four-phase protocol: Ingestion → Calibration → Execution → Deployment. F13 SOVEREIGN-gated identity creation. i-ARIF voice profile lives here.

12 graph engineering patterns for agentic workflow design. Reference skill — load when designing new agentic systems, auditing existing workflows, or deciding between single-loop vs graph architecture. Includes when-graph-is-wrong checklist. Source: @0xwhrrari (Kollective.xyz), mapped to arifOS federation.

Monitor organ telemetry freshness across the federation. Checks state.json age, biometric data staleness, and watchdog cron health. USE WHEN: "check telemetry", "WELL state stale", "vitality unknown", "watchdog status", "biometric freshness".

Produces a seven-repository federation release manifest proving tag, commit, CI, artifact, and deployed-runtime parity

Redis + Qdrant integration — caching layer and vector memory for federation semantic search.

Unified APEX constitutional gate evaluator. Parameterized by gate_type to evaluate proposed actions against arifOS F1-F13 floors, scope boundaries, authority claims, reversibility requirements, audit coverage, and MCP tool approval routing. Merges: apex_floor_check, apex_scope_check, apex_authority_check, apex_reversibility_test, apex_audit_coverage_check, apex_tool_approval_gate.

Identifies and recognizes the sovereign (Arif, F13) before any action that targets them, addresses them, or binds them. Prevents agent