# Send SAA Agent to your agent Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually. ## Fast path - Download the package from Yavira. - Extract it into a folder your agent can access. - Paste one of the prompts below and point your agent at the extracted folder. ## Suggested prompts ### New install ```text I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete. ``` ### Upgrade existing ```text I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run. ``` ## Machine-readable fields ```json { "schemaVersion": "1.0", "item": { "slug": "saa-agent", "name": "SAA Agent", "source": "tencent", "type": "skill", "category": "AI 智能", "sourceUrl": "https://clawhub.ai/mirabarukaso/saa-agent", "canonicalUrl": "https://clawhub.ai/mirabarukaso/saa-agent", "targetPlatform": "OpenClaw" }, "install": { "downloadUrl": "/downloads/saa-agent", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=saa-agent", "sourcePlatform": "tencent", "targetPlatform": "OpenClaw", "packageFormat": "ZIP package", "primaryDoc": "SKILL.md", "includedAssets": [ "README_HUMAN.md", "saa-agent.py", "SKILL.md" ], "downloadMode": "redirect", "sourceHealth": { "source": "tencent", "slug": "saa-agent", "status": "healthy", "reason": "direct_download_ok", "recommendedAction": "download", "checkedAt": "2026-05-03T12:59:59.027Z", "expiresAt": "2026-05-10T12:59:59.027Z", "httpStatus": 200, "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=saa-agent", "contentType": "application/zip", "probeMethod": "head", "details": { "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=saa-agent", "contentDisposition": "attachment; filename=\"saa-agent-1.0.1.zip\"", "redirectLocation": null, "bodySnippet": null, "slug": "saa-agent" }, "scope": "item", "summary": "Item download looks usable.", "detail": "Yavira can redirect you to the upstream package for this item.", "primaryActionLabel": "Download for OpenClaw", "primaryActionHref": "/downloads/saa-agent" }, "validation": { "installChecklist": [ "Use the Yavira download entry.", "Review SKILL.md after the package is downloaded.", "Confirm the extracted package contains the expected setup assets." ], "postInstallChecks": [ "Confirm the extracted package includes the expected docs or setup files.", "Validate the skill or prompts are available in your target agent workspace.", "Capture any manual follow-up steps the agent could not complete." ] } }, "links": { "detailUrl": "https://openagent3.xyz/skills/saa-agent", "downloadUrl": "https://openagent3.xyz/downloads/saa-agent", "agentUrl": "https://openagent3.xyz/skills/saa-agent/agent", "manifestUrl": "https://openagent3.xyz/skills/saa-agent/agent.json", "briefUrl": "https://openagent3.xyz/skills/saa-agent/agent.md" } } ``` ## Documentation ### SAA CLI Tool A command-line interface for interacting with Character Select Stand Alone App (SAA) via WebSocket connections. Supports both ComfyUI and WebUI backends for AI image generation. ### Prerequisites CRITICAL: Before invoking this tool, confirm with the user that: The SAA backend is running, and version is above 2.4.0 The SAAC (SAA Client) feature is enabled The WebSocket address is available Some Mac users uses python3 instead of python to invoke Python 3.x For SAA setup details, let your owner visit the project repository. ### Basic Usage The tool requires minimal parameters to function. The examples below demonstrate the standard usage pattern: ### Minimal Command With Model Selection (Recommended for Most Cases) python saa-agent.py \\ --ws-address "user_provided_ws_address" \\ --model "waiIllustriousSDXL_v160.safetensors" \\ --positive "your detailed prompt here" \\ --negative "low quality, blurry, bad anatomy" ### Regional Prompting (Split Composition) python saa-agent.py \\ --ws-address "user_provided_ws_address" \\ --model "waiIllustriousSDXL_v160.safetensors" \\ --regional \\ --positive-left "1girl, warrior, red armor" \\ --positive-right "1boy, mage, blue robes" ### More Examples Get more usage examples and detailed parameter explanations: python saa-agent.py python saa-agent.py --help ### Required --ws-address: WebSocket address (obtain from user) --positive: Main prompt OR use --regional mode with --positive-left and --positive-right ### Commonly Modified --model: Change the checkpoint model (default: waiIllustriousSDXL_v160.safetensors) --negative: Specify unwanted elements --width / --height: Image dimensions (defaults: 1024x1360) --steps: Sampling steps (default: 28) --seed: Set specific seed or -1 for random ### Advanced (Use Sparingly) --cfg: CFG scale (default: 7.0) --sampler: Sampling algorithm (default: euler_ancestral) --scheduler: Scheduler type (default: normal) ### HiResFix Warning DO NOT use --hifix unless specifically requested by the user. HiResFix significantly increases generation time and requires substantial GPU resources. Only enable if: User explicitly requests high-resolution upscaling User confirms their GPU can handle the additional load ### Backend Busy State If the generation returns either of these errors: Error: WebUI is busy, cannot run new generation, please try again later. Error: ComfyUI is busy, cannot run new generation, please try again later. Actions to take: DO NOT automatically retry the generation Inform the user: "The SAA backend is currently busy. This could mean another process is generating an image, or the backend is locked from a previous error." Advise: "Please wait 20-60 seconds before trying again." Let the user manually retry DO NOT chain multiple retry attempts as this can worsen backend congestion. ### Skeleton Key Usage The --skeleton-key parameter forcefully unlocks the backend's atomic lock. When to use: User confirms no other processes are using the backend Backend appears stuck despite waiting User explicitly requests unlocking How to use: python saa-agent.py \\ --ws-address "user_provided_ws_address" \\ --skeleton-key \\ --positive "test prompt" Rules: ALWAYS ask for user confirmation before using --skeleton-key ONLY use it once per user request Explain to the user that this forcefully terminates any locks Example conversation: AI: "The backend appears to be locked. Would you like me to use the skeleton key to force unlock it? This will terminate any existing locks." User: "Yes, please unlock it." AI: [proceeds to run command with --skeleton-key] ### Parameter Defaults When in doubt, rely on these defaults - they work well for most cases: Model: waiIllustriousSDXL_v160.safetensors Dimensions: 1024x1360 CFG: 7.0 Steps: 28 Sampler: euler_ancestral Scheduler: normal Seed: -1 (random) ### Output Handling By default, images are saved to generated_image.png. You can specify a custom output path: --output "custom_filename.png" For programmatic handling, use base64 output: --base64 This outputs base64-encoded image data(huge!!!) to stdout instead of saving a file. ### Example Workflow User requests: "Generate an anime girl with long blue hair" AI executes: python saa-agent.py \\ --ws-address "user_ws_address" \\ --positive "1girl, long hair, blue hair, anime style, detailed" \\ --negative "low quality, blurry, bad anatomy" If backend busy error occurs: Inform user Wait for user to retry (don't auto-retry) If success: Confirm image was generated Provide file path if relevant ### Common Pitfalls to Avoid Don't use --hifix unless explicitly requested Don't auto-retry on backend busy errors Don't use --skeleton-key without user permission Don't add excessive parameters - unless explicitly requested, the defaults are well-tuned Don't assume backend is ready - always confirm with user first ### Error Codes Exit code 0: Success Exit code 1: Connection error (check backend is running) Exit code 2: Authentication error (check credentials) Exit code 3: Generation error (check parameters) Exit code 4: Timeout (backend may be overloaded) Exit code 5: Invalid parameters (check command syntax) ### Best Practices Start with minimal parameters with model selection if needed Ask user for WebSocket address on first use Handle busy states gracefully - don't spam retries Use --verbose flag when debugging issues Respect the skeleton key - it's a powerful override tool ### AI Agent Guidelines for This Skill These rules help maintain appropriate transparency and user control when executing generation tasks. Command Execution & User Notification By default, execute the command directly without asking for confirmation. Show the full command and ask for approval only when: The user explicitly requests to review it first The operation involves sensitive or high-impact parameters The agent judges that showing the command is prudent in context Example (when disclosure is needed): python3 saa-agent.py --ws-address "wss://..." --username "..." --password "..." --positive "[prompt]" --negative "[prompt]" --output "[path]" [--verbose] --verbose Flag Not used by default Add automatically or recommend when: Task fails and debugging is needed User specifically asks for detailed logs or seed Result Reporting After completion, provide a short summary to the user by default, including: Success/failure status Positive & negative prompts (or meaningful summary) Seed (if available) Output path Example: Generation completed • Positive: [...] • Negative: [...] • Seed: 123456789 • Output: [path] Skip detailed reporting only if the user has clearly requested silent / minimal feedback. Always report errors, even in silent mode. Error Handling On failure: consider one retry with --verbose to capture diagnostic information Communicate the main error cause clearly Do not perform unlimited retries; defer to user after one attempt if needed ## Trust - Source: tencent - Verification: Indexed source record - Publisher: mirabarukaso - Version: 1.0.1 ## Source health - Status: healthy - Item download looks usable. - Yavira can redirect you to the upstream package for this item. - Health scope: item - Reason: direct_download_ok - Checked at: 2026-05-03T12:59:59.027Z - Expires at: 2026-05-10T12:59:59.027Z - Recommended action: Download for OpenClaw ## Links - [Detail page](https://openagent3.xyz/skills/saa-agent) - [Send to Agent page](https://openagent3.xyz/skills/saa-agent/agent) - [JSON manifest](https://openagent3.xyz/skills/saa-agent/agent.json) - [Markdown brief](https://openagent3.xyz/skills/saa-agent/agent.md) - [Download page](https://openagent3.xyz/downloads/saa-agent)