Openclaw Workspace

Grok documentation and configuration help. Use when users ask about setup, configuration, MCP servers, authentication, skills, slash commands, keyboard shortcuts, or any Grok feature. Also use proactively when you detect a user is having trouble with setup or onboarding.

runpod-flash — code-first serverless: write Python locally, run it on remote Runpod GPUs/CPUs with `flash dev` (hot-reload + live worker logs), then `flash deploy`. Use for @Endpoint/@remote functions, resource config, and debugging flash deployments. For CLI-only infra management use runpodctl or runpod-mcp.

Start here for any Runpod task — running GPU/CPU pods, deploying serverless endpoints, templates, network volumes, building images, or understanding how Runpod works. Routes the request to the right Runpod skill (runpod-mcp, runpodctl, flash, companion-clis, or runpod-usage). Use when it is unclear which Runpod skill applies.

How to use the image_gen and image_edit tool calls in Grok Build: when to build a visual with code instead of generating it, prompt-craft, reference-first handling of real people, factual grounding, and asset-consistency. Load this whenever generating or editing an image is on the table, i.e. when an image_gen or image_edit call is being considered or about to be made. Tool-usage-driven, not triggered by a user merely mentioning images.

Use mmx to generate text, images, video, and speech via the MiniMax AI platform. Use when the user wants to create media content, chat with MiniMax models, perform web search, or manage MiniMax API resources from the terminal.

OpenClaw edge agent bridge — operational triage, doctor, restart, and A2A bridge routing for the federation edge (Telegram surface). USE WHEN: "openclaw unhealthy", "gateway down", "edge bot not responding", "a2a bridge disconnected", "watchdog tripped", "openclaw doctor", "openclaw restart". NOT for token/security audit — use FORGE-telegram-audit.

Runpod CLI for managing GPU/CPU workloads from the terminal — pods, serverless endpoints, templates, network volumes, Hub deploys, models, SSH, and file transfer (send/receive). Use for terminal/CI/scripting, Hub browse/deploy, SSH setup, `doctor`, or when the Runpod MCP tools are not connected. For structured tool calls in an MCP-enabled session, prefer runpod-mcp.

FEDERATED Xiaomi MiMo v2.5 audio intelligence — TTS (0-credit free window), ASR (en/zh), voice design, voice clone (F13-gated). Available to ALL AAA agents via this canonical skill. Use when generating speech/voice notes, designing voices, or transcribing wav/mp3 (en/zh). BM/dialect stays on GLM/dashscope lanes. TRANSPORT: direct lane (litellm cannot carry audio bodies).

Manage Runpod infrastructure — pods, serverless endpoints, jobs, templates, network volumes, container-registry auth, GPU/CPU catalog, and billing — via the Runpod MCP server's structured tool calls. Use when the Runpod MCP tools (create-pod, list-endpoints, …) are connected in this session, or to connect them (hosted OAuth or local npx). Prefer this over runpodctl for plain infra CRUD when MCP is available; use runpodctl for the terminal, file transfer, or SSH setup.

Build multi-step execution graphs, dependency-aware subtasks, checkpoints, and rollback points. Load when tasks exceed one-shot prompting and need subagent or staged execution.

Probe Z.AI GLM coding plan health — models list, anthropic + openai endpoints, credit reality, cross-check 5 wired surfaces. Use when Arif says 'zai probe', 'check Z.AI plan', 'is GLM alive', 'zai health', or when any zai-backed harness (Qwen/Kimi/OpenCode/Claude) behaves oddly.

When user asks to summarize text, articles, documents, meetings, emails, YouTube transcripts, books, PDFs, reports, conversations, or any long content

Probe all FI coder CLIs live (qwen/kimi/opencode/codex/claude/grok/gemini) with a minimal falsification prompt and report the mesh matrix. Use when Arif says 'mesh check', 'test all coder CLIs', 'FI mesh health', 'are all harnesses alive', or before/after cross-FI plumbing work.

Build, test, inspect, install, and deploy MCP servers with FastMCP in Python

Build and audit interactive MCP Apps, UI resources, sandbox messaging, and CSP-bound host integration

Guide the sovereign or operator through any setup, configuration, or deployment process step by step. One atomic action per response, with a 'Still remaining' list that never exceeds 8 items. Use when Arif says 'help me set up', 'walk me through', 'setup-help', 'configure X', or any guided process that requires tracking remaining steps across multiple turns.

Companion CLIs for Runpod workflows — HuggingFace, GitHub, Docker, and AWS.

Use when mapping every callable in the federation or planning boot-time call contracts between FI harnesses. Use when mapping every callable in the federation or planning boot-time call contracts between FI harnesses. Boot-time call contract: how to invoke every FI harness, organ MCP, A2A target, and FED socket. AAA 3-layer cards = directory (who). This skill = telephone (how). Load on session start.

Meta-mesa orchestrator for Qwen Code (FI-003). Given a multi-step mission, decompose into L1/L2 skills, sequence them, route to the right organ (arifOS/A-FORGE/GEOX/WEALTH/WELL). Load this when a mission spans more than 2 skills or 3 steps, or when you feel yourself wandering between tools.

[QwenCloud] Generate text, have conversations, write code, reason, and call functions with Qwen models. TRIGGER when: user asks to chat with Qwen, generate text, write code with Qwen, use Qwen function calling, or explicitly invokes this skill by name (e.g. use qwencloud-text). DO NOT TRIGGER when: general coding questions without Qwen, non-Qwen AI model usage (OpenAI, Gemini, etc.), image/video understanding (use qwencloud-vision), image/video/audio generation.

Federation-wide gold (XAUUSD) trading capability. Python stack, OANDA broker, backtesting, macro signals, RSI strategy. Every organ has a role.

Shadow mode multimodal compute & generation doctrine for sovereign human needs. Governs out-of-band GPU execution (A100/Hostinger/RunPod), asynchronous batch processing, privacy isolation, and zero-clutter offloading.

Atomic-swap upgrade of the Qwen Code release install (SHA256-verified, rollback-preserved). Use when Arif says 'upgrade qwen', 'qwen update', 'new qwen version', or when `qwen update` refuses with a misleading git-clone message.

Configure, audit, and align Kimi Code CLI as AAA warga FI-008 with arifOS kernel and A-FORGE stdio actuator.

Orthogonal axis routing for Qwen Code (FI-003). Given an intent, classify it across 9 orthogonal axes (organ, action, mode, tier, layer, surface, scope, time, reversibility) and route to the right organ MCP server. Load when you must pick between equally-valid tools.

[QwenCloud] Manage account auth and query usage/billing. Use for: login, logout, check usage, view billing, free tier quota, coding plan status, pay-as-you-go costs. Skip for: model browsing, non-account tasks.

Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output should be a bitmap asset rather than repo-native code or vector. Do not use when the task is better handled by editing existing SVG/vector/code-native assets, extending an established icon or logo system, or building the visual directly in HTML/CSS/canvas.

Extend the arifOS dream-engine so every AAA warga (333-AGI, 555-ASI, 888-APEX, A-AUDIT, A-ARCHIVE), OpenCode, and OpenClaw can autonomously consolidate memory without violating F1-F13.

Meta-mesa orchestrator for Claude Code (FI-002). Given a multi-step mission, decompose into L1/L2 skills, sequence them, route to the right organ (arifOS/A-FORGE/GEOX/WEALTH/WELL). Load this when a mission spans more than 2 skills or 3 steps, or when you feel yourself wandering between tools.

Runbook for GitHub & Git operations across federation repos — commit workflow, PR ops, branch discipline. v1.1.0: organ paths from organ registry, no hardcoded /root/<repo> paths.

Standard procedure for registering a new agent in the AAA federation. Creates agent identity directory, agent card (v2.0.0 schema), registry entry, and SOUL.md. Updated 2026-08-03: canonical card locations vary by agent class — identity lanes at agent-cards/identity/, organs at agent-cards/{pillars,organs}/, FI agents (warga-aaa forge instruments, FI-001 to FI-008) at agents/_external/<id>/, other internal agents at agents/<id>/. The 2026-07-01 single-path note is now obsolete.

Generate, monitor, and download MiniMax-H3 videos through the mmx CLI. Use for H3 text-to-video, first/last-frame video, multimodal reference image/video/audio generation, H3 prompt improvement, media preflight, Pay-as-you-go API key selection, task waiting, downloads, and H3 failure handling.

Route all Malaysia macro/policy/energy/corporate claims through primary government/financial sources using EXISTING federation tools. NO new MCP servers. Hard F2 provenance law for any MY figure.

One-click deploy, publish, and update a local project or Git repository to Alibaba Cloud International (alibabacloud.com), producing an accessible online service with a public IP. Supports full-stack ROS orchestration, automatic cloud-resource provisioning, pre-deployment price confirmation, service health checks, deployment-state recording, hot updates, and optional domain + HTTPS setup. Use when: the user asks to deploy a project to the cloud, put an app online, publish a website, generate an access URL, deploy a Git repo, or update an online version and has NOT named a specific cloud platform; or the user mentions "Alibaba Cloud", "alibabacloud.com", or the international site. Do not use when: the user explicitly targets Aliyun China (aliyun.com), AWS, GCP, Azure, or another specific cloud platform.

[QwenCloud] Understand images and videos with Qwen vision models. TRIGGER when: user wants to analyze, describe, or extract information from images or videos, OCR text extraction, chart/table reading, visual reasoning, multi-image comparison, screenshot understanding, video comprehension, or explicitly invokes this skill by name (e.g. use qwencloud-vision). DO NOT TRIGGER when: user wants to generate/create images (use qwencloud-image-generation), generate videos (use qwencloud-video-generation), text-only tasks without visual input, or non-Qwen vision tasks.

Call Qwen Token Plan image models on the Personal allowlist (qwen-image-3.0-pro, wan2.7-image, wan2.7-image-pro). Activate when the user asks to draw, generate, or edit an image via Token Plan.

Call Qwen Token Plan video models on the Personal allowlist (happyhorse-1.1-t2v / i2v / r2v). Activate for text-to-video, image-to-video, or reference-to-video on Token Plan.

Orthogonal axis routing for Claude Code (FI-002). Given an intent, classify it across 9 orthogonal axes (organ, action, mode, tier, layer, surface, scope, time, reversibility) and route to the right organ MCP server. Load when you must pick between equally-valid tools.

Parse failing GitHub Actions logs, identify root cause patterns, and propose fixes without executing irreversible changes. Use this skill whenever a federation repo shows a red CI status, a workflow fails, or a build/test/lint gate breaks. This skill reads logs, classifies failure modes, and outputs a diagnostic report — it does not re-run CI, edit workflows, or dismiss security findings without sovereign approval.

AI-powered web scraping and extraction. ScrapeGraph AI for structured extraction with schema generation and async crawling. Decodo for high-success web scraping across 195+ locations with residential proxies, anti-bot handling, eCommerce, social media, and AI search integration.

Self-hosted + commercial web search with routing logic. Private SearXNG search, multi-engine SerpApi extraction, and Firecrawl web search + scrape + interact + parse + monitor + research. Routes to the right search backend based on task.

Generate images, videos, TTS, voice clone, and music via MiniMax MCP server. Use when user asks to "draw", "generate image", "create picture", "make a photo", "text to image", "image generation".

Call Qwen Token Plan speech models on the Personal allowlist (TTS, ASR, realtime). Activate for text-to-speech, speech-to-text, or live voice on Token Plan.

Standardized JSON-RPC state machine, task queue delegation, and precondition contract manager for cross-vendor multi-agent swarms.

ASR (Automatic Speech Recognition) ingestion layer using GLM-ASR-2512 (Z.AI). Decode phase of AAA-audio-emd-pipeline. Custom dictionary for arifOS-specific terms, audio chunking for ≤30 s/≤25 MB constraint, stream mode for realtime. Penang-Besi dialect and code-switching support.

REPO_REALITY_AUDIT::v1.2 — Sovereign doctrine (F13, 2026-08-14). Audit codebase reality, not AI output. Categories: STUB Tiers T1-T4, ORPHAN, SHIM, REALITY_LEAK, AUTHORITY_LEAK, CLAIM_DRIFT (A comment/B architecture/C metric), FAKE_METRIC. Pipeline: SCAN→CONTEXT→REACHABILITY→RUNTIME_IMPACT→VERDICT. HEURISTIC≠MEASUREMENT semantics. Nuclear Rule: no abstract feature without concrete consumer. USE WHEN: audit repo, find stubs, reality audit, orphan/shim detection, verify implementation claims.

Hermes cognitive slash commands — /NNN_word format. 8 numerological spine, 16 cognitive verbs. Zen layer for arifOS federation.

Governed triage workflow for GitHub issues across the arifOS federation Use this skill whenever a new issue is opened, an issue lacks labels, or an agent needs to determine if it belongs in a different repo or federation organ. This skill classifies, routes, labels, and drafts responses — but never closes, never assigns to Arif, and never promises code fixes without sovereign approval.

Close remaining Seal-A gates after P0 proof/identity/ACT/T3a work. Use when: Seal-A, stage 000, SE stage engine, SOT v2, BOOT, T3a binding, free_nonce, constitutional_grade, remaining seal tasks.

Closed-loop visual governance with W3 tri-witness consensus — render, screenshot, validate, iterate, seal. Includes MCP App / ChatGPT host-surface QA.

Persistent agentic state doctrine for Qwen Code (FI-003). Survive across sessions by writing to /root/.qwen/agent_state/. Auto-memory (MEMORY.md) holds lessons learned; this holds session-overlapping capability + mission history. Load when you want to accumulate intelligence across sessions.

Documents Qwen Token Plan Harness tools (web search, code interpreter, web scraping, reverse image search, text-to-image search) that are built into supported models. Activates when the user asks for real-time web info, code execution in a sandbox, or image search while using a Token Plan model.

[QwenCloud] Configure authentication (API keys, endpoints). TRIGGER when: setting up QWEN_API_KEY, troubleshooting 401/auth errors, when another skill reports missing credentials, or user explicitly invokes this skill by name (e.g. use qwencloud-ops-auth). DO NOT TRIGGER when: non-auth Qwen tasks, general API usage questions.

Use when the user asks how to build with OpenAI products or APIs, asks about Codex itself or choosing Codex surfaces, needs up-to-date official documentation with citations, help choosing the latest model for a use case, latest/current/default-model prompting guidance, or model upgrade and prompt-upgrade guidance; use OpenAI docs MCP tools for non-Codex docs questions, use the Codex manual helper first for broad Codex self-knowledge, and restrict fallback browsing to official OpenAI domains.

Eureka, contradiction, insight generation, and cross-domain synthesis for AAA warga

Unified MCP operations skill. General MCP server operations via mcporter CLI, federation MCP management (FastMCP build/deploy/operate), and MCP server health probe + auto-recovery (lifeguard). Merges: FORGE-mcp-ops, FORGE-mcp-federation-ops, FORGE-mcp-lifeguard.

Docker Compose runbook for af-forge VPS. Pinned versions, restart discipline, health checks, compose governance, thermodynamic fleet reasoning, and concrete container commands. Lower machine entropy.

This skill should be used when the user needs to analyze Git repositories, compare developer commit patterns, work habits, development efficiency, code style, code quality, and slacking behaviors. It generates honest, direct developer evaluations with scores, grades, strengths, weaknesses, and actionable suggestions. Trigger phrases include "analyze code", "analyze repository", "compare developers", "code quality report", "commit patterns", "developer efficiency", "developer evaluation", "slacking index", "摸鱼指数", "工作习惯分析", "代码分析", "研发效率", "代码质量", "开发者评估", "developer score".

Monitor the model fallback chain. Track latency, billing failures (402), cold-start failures, and auto-pause dead models. USE WHEN: "model health", "check fallback chain", "model latency", "billing alert".

Federation-wide broken symlink scanner. Scans /root for broken symlinks, categorizes by location, and reports with safe-delete recommendations. USE WHEN: "check symlinks", "broken links", "symlink debt", "find broken symlinks", or during entropy sweeps.