Openclaw Workspace

All computation is mathematical. Uncertainty is quantified. Optimization has structure. Proof has rules. Numbers don't lie but models can.

[QwenCloud] Synthesize speech from text with Qwen TTS models. TRIGGER when: user wants to convert text to speech, create voiceovers, generate audio narration, read text aloud, build TTS applications, mentions speech synthesis/voice generation/audio output from text, or explicitly invokes this skill by name (e.g. use qwencloud-audio-tts). DO NOT TRIGGER when: user wants speech recognition/ASR, text generation without audio, non-Qwen audio tasks.

Unified recursive-improvement protocol across ALL agents in the federation (AAA, Hermes, Kimi, OpenClaw, Claude Code, Codex, OpenCode, Grok). Bridges session-level RSI to federation-wide sync: detects skill drift between agents, propagates fixes back to the canonical AAA catalog, and ensures the federation learns from every session — not just the agent that ran it.

Perform a read-only, defect-first review of a specified code change and return every actionable finding. Use when another agent delegates review of uncommitted changes, a base-branch diff, a commit, or custom review instructions.

Make web content maximally extractable by LLMs, Agentic RAG systems, and search indexers.

Use when initializing any agent, designing governance, or establishing compact operating constitution. Use when initializing any agent, designing governance, or establishing compact operating constitution. Compact operating constitution for every AAA agent. Load before any non-trivial action. Distills 10 Agent Invariants + 12 governance rules + skills audit into portable doctrine. Covers tool classification, evidence/authority separation, degradation dominance, propose-before-execute, and memory atoms.

Persistent agentic state doctrine for Claude Code (FI-002). Survive across sessions by writing to /root/.claude/agent_state/. Auto-memory (CLAUDE_IDENTITY.md + memory/) holds lessons learned; this holds session-overlapping capability + mission history. Load when you want to accumulate intelligence across sessions.

Validate Caddy reverse proxy configs, Cloudflare Origin CA SSL expiry, DNS parity, tunnel status, and Cloudflare tunnel exposure. USE WHEN: "check SSL", "Caddy config valid", "tunnel status", "DNS parity", "SSL expiry", "infra health", "Cloudflare check".

Next.js App Router engineering — server components, streaming, and edge patterns.

Audit env.local, SOPS .env, and config files for plaintext secret leaks, key age, missing rotation dates, and overlong-lived credentials. USE WHEN: "secret audit", "key age", "rotate secrets", "credential hygiene", "API key check", "env audit", "secret leak scan".

Standard contract for spawning bounded, auditable sub-agents — input contract, output schema, evidence requirements.

Automated TREE777 security checks for Telegram bot tokens, webhook isolation, bot permission scope, and A2A bridge security. v1.1.0: organ paths from registry, ports from live probes. USE WHEN: "telegram audit", "bot security", "webhook check", "token isolation", "TREE777 check", "telegram permissions".

Governance layer for image generation that bridges the dimensionality deficit between sparse human prompts and high-dimensional pixel output. USE WHEN: "generate image from text prompt", "T2I dispatch", "diffusion call", "image generation", "vision tool call". Implements the immutable JSON receipt contract (f1_safe, f2_adherence, prompt_density, hallucinated_elements, anchor_required, anchor_suggestion) with hard/soft ΔS gate thresholds (0.20, 0.50) and the Hybrid heuristic-plus-VLM density estimator.

Access thousands of curated AI system prompts, role personas, task templates, and variable substitution templates directly via prompts.chat MCP server (@fkadev/prompts.chat-mcp).

Use when sealing audit decisions or recording constitutional events. Every decision logged. Irreversible decisions sealed. ΔS ≤ 0 on every output. Receipts > narratives.

Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Codex's capabilities with specialized knowledge, workflows, or tool integrations.

Use after task completion for lightweight post-task reflection — list key decisions made under uncertainty. Use after task completion for lightweight post-task reflection — list key decisions made under uncertainty. Use after task completion for lightweight post-task reflection — list key decisions made under uncertainty. Lightweight post-task reflection: list key decisions made during work that the agent is uncertain about. Triggered manually by Arif ('/decisions', 'what are you unsure about', 'review your logic') or automatically after major refactors, multi-file changes, or SEAL-grade work. Complements QQQ protocol (governed decisions) with quick uncertainty surfacing.

Multi-modal reasoning bridge — unifies text, image, tabular data, and geospatial evidence into a single reasoning chain. Routes visual analysis through GEOX vision pipeline, tabular data through WEALTH compute, and text through standard LLM reasoning. USE WHEN: "analyze this image + data", "combine visual and textual evidence", "multi-modal prospect evaluation".

Komda Color Law lint — enforces SOVEREIGN_DECREES §04 (territory color families). Use when generating, editing, or auditing any visual artifact (HTML/SVG/CSS) that claims a federation territory (arifos / geox / wealth / well / aaa). Foreign-family color in territory DOM = violation per F13 SOVEREIGN doctrine (2026-08-01).

MCP + A2A protocol integration — agentic inter-agent communication and task delegation.

Tailwind design tokens — Trinity Design System for federation cockpit and web surfaces.

Governed multi-CLI coding fabric. Hermes dispatcher routes intent to specialist workers (qwen / claude / codex / kimi / gh). F13 enforcement via wrapper executables, not prompt-only. Hermes accepts only verifiable artifacts (diff, test result, lint, commit SHA, PR URL). First milestone of arifOS coding-fabric doctrine.

Adds visual understanding to text-only Token Plan models (glm-5, glm-4.7, MiniMax-M2.5, qwen3-max-2026-01-23, qwen3-coder-next, qwen3-coder-plus) by routing image inputs through a vision-capable model. Activates when the user passes an image to a text-only model or asks 'what's in this image'.

Operates the perpetual A-FORGE reality loop across observation, hypothesis, execution, entropy checks, and sealing. The 7-stage intent compiler that makes 000→999 self-sustaining. Auto-starts on complex work, auto-seals when zen marginal value → 0, carries forward to next session.

Bind governance before action. Every session must initialize constitutional floors, recognize sovereign signals, and classify autonomy tier before any other verb.

Five gates before commitment: authority + evidence + reversibility + lineage + REALITY. All five must open. One missing = HOLD.

Create and scaffold plugin directories for Codex with a required `.codex-plugin/plugin.json`, optional plugin folders/files, valid manifest defaults, and personal-marketplace entries by default. Use when Codex needs to create a new personal plugin, add optional plugin structure, generate or update marketplace entries for plugin ordering and availability metadata, or update an existing local plugin during development with the CLI-driven cachebuster and reinstall flow.

EMD (Encode → Metabolize → Decode) orchestration doctrine for audio in the arifOS federation. The constitutional loop that binds ASR → LLM reasoning → TTS into a single governed pipeline. Every audio exchange in arifOS flows through this reflex arc.

Translate internal federation and epistemic vocabulary into publication-grade geological prose before rendering human-facing PDF artifacts.

TTS engine registry for the arifOS federation. ChatTTS (qualitative conversational), F5-TTS / E2-TTS (zero-shot flow matching), Fish Speech (audio language model), MiniMax speech-2.8-hd (HD commercial), Edge / Mulberry (free fallback), MiMo Token Plan (Xiaomi multimodal subscription). Trade-off matrix, F13 governance, and routing rules. Single source of truth for which engine to use when.

Governed intelligence skill for AAA as the abstraction, attestation, and abduction control plane across arifOS, APEX, A-FORGE, GEOX, WEALTH, WELL, and the ariffazil profile repository. Use when the user asks to explain or design AAA, route agentic work, reduce chaos/entropy in an arifOS federation task, create AREP/task declarations, classify risk, plan multi-repo changes, review governance boundaries, or translate human intent into evidence-backed, authority-safe, recursively agentic workflows. Provides deterministic F1-F13 floor checking, bounded abduction, and FederationReceipt composition.

Decentralized identifier (did:web) identity management for federation organs and agents.

Operate Google Workspace from one CLI using dynamic API discovery, secure OAuth flows, and agent-ready automation patterns for Drive and

VAULT999 witness — immutable ledger integration, seal chain verification, and audit trace.

Use when modeling human behavior or intent.

Estimate human state State(t) = f(Energy, Attention, Optionality, Governance, Meaning, Witness), compute DesiredState(t+1), and select BestMinimalIntervention. Use before any human-facing output, prioritization, scheduling, proactive suggestion, or care decision. Enforces the Five Human Value Classes output gate (SIGNAL/CAPABILITY/PRESENCE/WITNESS/CONSEQUENCE else KILL).

All reality claims are physical claims. Conservation laws, thermodynamics, causality, entropy. You cannot get something from nothing.

Mapbox developer API suite, cartographic design, style management, GeoJSON validation/preview, coordinate conversion, and token security via Mapbox DevKit MCP server (mcp-devkit.mapbox.com).

OpenClaw integration for the human-meaning-membrane doctrine — apply the inference protocol when reviewing human-facing code (chatbots, UIs, social features). USE WHEN: "review this chatbot PR", "audit user-facing copy", "check consent logic", "message tone analysis code", "social feature review", "human-interaction code review". NOT for OpenClaw ops/health — use openclaw skill. NOT for token audit — use FORGE-telegram-audit.

[QwenCloud] Check for qwencloud-ai updates and notify the user when a new version is available. TRIGGER when: user asks to check for updates, check version, asks 'is there a new version', 'latest version', 'update skills', 'check update', or any other qwen skill delegates to this skill, or user explicitly invokes this skill by name (e.g. use qwencloud-update-check). DO NOT TRIGGER when: non-update-related tasks, general version questions about other software.

Install Codex skills into $CODEX_HOME/skills from a curated list or a GitHub repo path. Use when a user asks to list installable skills, install a curated skill, or install a skill from another repo (including private repos).

Cultural, dignity, and sovereignty lens for AAA state records. Apply before sealing decisions that touch identity, privacy, boundaries, or sovereign veto.

External constitutional audit and bounded forge planning for arifOS federation organs. Detects drift, challenges architecture claims, and prepares contracts.

Pattern for auditing the arifOS kernel when multiple MCP tools share the same skeleton but disagree on identity/verdict/affordance — proven

In-process columnar SQL analytics engine over Parquet, CSV, JSONL, and well log (.las) files using DuckDB and Polars.

Build GitHub Actions CI/CD pipelines, Docker multi-stage builds, and deploy-verify patterns for VPS af-forge. Cover test → build → deploy → smoke-test → rollback lifecycle.

Run an extremely strict maintainability review for abstraction quality, giant files, and spaghetti-condition growth. Standalone maintainability review focused on implementation quality, codebase health, and structural simplification.

Source-vs-runtime drift detection + live baseline comparison. Detect drift between federation source and runtime. Reads organ topology from canonical SOT, probes live health endpoints, compares git SHAs, and compares live state against saved baselines across tool manifests, agent cards, skill registries, and runtime-injected files.

Unified MCP server testing, probing, and conformance validation. Covers MCPJam Inspector methodology for deep conformance testing, MCPJam CLI/SDK probing for protocol version and stateless transport discovery, and quick smoke tests for federation organ health. Merges: FORGE-mcp-smoke-test, FORGE-mcp-probe, FORGE-mcp-testing.

Drift detection, skill binding, and federated architecture management. Real-time drift detection across tool manifests, agent cards, skill registries, and runtime-injected files. Discover, bind, and compose skills across all federation organs using orthogonal axes. Design, validate, and manage skills across a multi-agent federation with 3-layer architecture.

Verification-as-terminal-state doctrine + subagent verification tool. A task is done ONLY when verified. Never stop at "I changed it" — only at "it's fixed and confirmed." Runs health probes, behavior smoke tests, drift checks, and spawns verifier subagents.

When no permanent tool exists for a task, spawn a temporary tool via forge_ephemeral. The capability metabolism engine — generate, test, use, then dissolve. Prevents tool accumulation. Adapts without accumulating permanent state. Wolf Cabinet Ψ Survival — "I need to do X but no tool exists."

ESM/CommonJS interop guard - blocks silent require() failure of .mjs modules at pre-commit. Foodset-derived from scar-001-esm-sct-silent-fail (2026-08-13). Static analysis, read-only.

Governance JSON-LD — constitutional ontology and semantic governance context definitions.

MANDATORY LSP grounding gate BEFORE any code mutation on .ts, .py, .js, .tsx, .jsx files. Forces the agent to read real-time compiler diagnostics and structural project context before editing — eliminating blind guesses and anchoring every mutation in F2 (TRUTH). Routes through arifOS kernel (:8088) for centralized gate logic.

Federation-wide multimodal routing skill. Resolves any vision/audio/video/somatic request to its canonical model + endpoint via /root/.config/federation-models.json (SOT). USE WHEN: "which model should I use for X", "route this image/audio/video to the right engine", "unify multimodal routing across agents", "I have an image/audio/video and need to know where it goes".

Runtime for musyawarah (independent 333 ARCHITECT + 555 AUDITOR) then gotong-royong (sequential execute hop). Authority star, evidence as position files. Not a chatboard. Grok: workflow musyawarah-gotong.

Standard procedure for registering a new agent in the AAA federation Creates agent identity directory, agent card, registry entry, and SOUL.md.

SOLE controller skill for repository intelligence across the arifOS Federation. Exposes 12

Use when task may be over-engineered. Route to the SMALLEST capability that accomplishes it. Prevent yak-shaving. Use when task may be over-engineered. Route to the SMALLEST capability that accomplishes it. Prevent yak-shaving. Route every task to the SMALLEST capability that can accomplish it. Prevents over-engineering, yak-shaving, and reflexive tool escalation. Enforces one-owner-per-task, idempotent actions, and strict queues. Ask: "Can a simpler tool do this?" before reaching for the powerful one.