Results for “sub-agent”

Design structured evaluation rubrics for scoring LLM and agent outputs — defining quality dimensions, scoring scales, hard gates, score descriptions, and edge cases. Load when the user asks to create an eval rubric, define evaluation criteria, design scoring dimensions, write an eval spec, or says "what should I evaluate", "design a rubric", "create eval criteria", "define quality dimensions", "evaluation rubric for", "how do I measure quality of". Sub-skill of eval-output orchestrator.

AI-powered crypto trading agent, wallet API, and LLM gateway via natural language. Use when the user wants to trade crypto, check portfolio balances (with PnL and NFTs), view token prices, search tokens, transfer crypto, manage NFTs, use leverage, bet on Polymarket, deploy tokens, set up automated trading, sign and submit raw transactions, or access LLM models through the Bankr LLM gateway funded by your Bankr wallet. Supports Base, Ethereum, Polygon, Solana, and Unichain.

Review the changes since a fixed point (commit, branch, tag, or merge-base) along two axes — Standards (does the code follow this repo's documented coding standards?) and Spec (does the code match what the originating issue/PRD asked for?). Runs both reviews in parallel sub-agents and reports them side by side. Use when the user wants to review a branch, a PR, work-in-progress changes, or asks to "review since X".

Create structured plans for multi-step tasks -- software features, research workflows, events, study plans, or any goal that benefits from breakdown. Also deepens existing plans with interactive sub-agent review. Use when the user says 'plan this', 'create a plan', 'how should we build', 'break this down', or when a brainstorm doc is ready for planning. Use 'deepen the plan' or 'deepening pass' for the deepening flow. For exploratory requests, prefer ce-brainstorm first.

Legacy-project style inheritance skill. Use when the user types /inherit-legacy-style, or when onboarding an AI coding agent onto a hand-written legacy project and you need to prevent "style drift" (the model imposing its pretrained mainstream idioms onto the project). Language- and framework-agnostic — it aligns meta-architecture only, not syntax. Once run, it becomes a behavioral constraint on all subsequent coding tasks. Do NOT use for pure research or one-off questions unrelated to code-style alignment.

Score LLM and agent outputs using LLM-as-judge techniques — direct scoring against rubrics or pairwise comparison between two outputs. Includes built-in bias mitigation for position bias, length bias, and self-enhancement bias. Load when the user asks to score an output, judge a response, evaluate against a rubric, compare two outputs, do direct scoring, run pairwise comparison, or says "rate this", "which response is better", "score this against the rubric", "judge this output", "LLM as judge this". Sub-skill of eval-output orchestrator.

Fully autonomous QA testing agent for Modular RAG MCP Server. Reads test cases from QA_TEST_PLAN.md, executes ALL test types automatically without human intervention — CLI commands, Dashboard UI via Streamlit AppTest headless rendering, MCP protocol via subprocess JSON-RPC, provider switches, and data lifecycle checks. Diagnoses failures, applies fixes with up to 3 retry rounds, and records results in QA_TEST_PROGRESS.md. Use when user says 'run QA', 'QA test', 'QA 测试', '执行测试', '跑测试', 'test and fix', or wants to execute QA test plan.

Fully autonomous QA testing agent for Modular RAG MCP Server. Reads test cases from QA_TEST_PLAN.md, executes ALL test types automatically without human intervention — CLI commands, Dashboard UI via Streamlit AppTest headless rendering, MCP protocol via subprocess JSON-RPC, provider switches, and data lifecycle checks. Diagnoses failures, applies fixes with up to 3 retry rounds, and records results in QA_TEST_PROGRESS.md. Use when user says 'run QA', 'QA test', 'QA 测试', '执行测试', '跑测试', 'test and fix', or wants to execute QA test plan.

Fully autonomous QA testing agent for Modular RAG MCP Server. Reads test cases from QA_TEST_PLAN.md, executes ALL test types automatically without human intervention — CLI commands, Dashboard UI via Streamlit AppTest headless rendering, MCP protocol via subprocess JSON-RPC, provider switches, and data lifecycle checks. Diagnoses failures, applies fixes with up to 3 retry rounds, and records results in QA_TEST_PROGRESS.md. Use when user says 'run QA', 'QA test', 'QA 测试', '执行测试', '跑测试', 'test and fix', or wants to execute QA test plan.

Capture actionable learnings that emerge during conversation — when the agent or user discovers that a skill, a set of skills, or a process needs to be updated based on what's happening in the current chat. Sub-skill of the learn-from orchestrator. Load when the user says "we should update the skill for this", "this should be a skill rule", "add this as a gotcha", "the skill should know about this", "update the process for this", "remember this for next time", "this is important for the skill". Also triggers when the agent notices a skill's guidance was wrong or incomplete, a process step failed or was unnecessary, a new pattern emerged, a guardrail was missing, a workaround became a pattern, or a debugging session reveals a gap.

Office-document renderers for pursuit deliverables — Markdown to DOCX (Pandoc/OpenXML) and JSON envelopes to styled XLSX (openpyxl). USE WHEN the user asks to export a Studio markdown file to Word, convert compliance matrix JSON to Excel, render proposal outline as DOCX, or run a one-off format conversion on files under pursuits/. Consumer skills (proposal-generator, subcontractor-sow-builder, compliance-auditor) call these scripts internally; users can also run renderers directly from Agent Skills or chat. DO NOT USE FOR drafting content (use proposal-generator), visual decks/PDF/PPTX (use huashu-design), or domain analysis.

Autonomously improve a real artifact (code, training recipe, agent harness, data pipeline, prompt) against an objective and an evaluator, using Hypothesis Tree Refinement (HTR) from the Arbor paper. Use this whenever someone wants to iteratively optimize something over many experiments without overfitting — e.g. "get my model's eval score up", "improve this agent/harness", "tune this pipeline", "beat the baseline on this benchmark", "run a search over approaches and keep the best", "do an MLE-bench / Kaggle-style optimization", or any long-horizon "make this artifact better and don't just memorize the dev set" task. Trigger it even when the user doesn't say "Arbor" or "hypothesis tree" but describes repeated experiment-and-evaluate loops, branching exploration of competing ideas, or worries about a dev/test gap. Runs Claude itself as the coordinator with subagent executors in...

Reduce OpenClaw token usage and API costs through smart model routing, heartbeat optimization, budget tracking, and native 2026.2.15 features (session pruning, bootstrap size limits, cache TTL alignment). Use when token costs are high, API rate limits are being hit, or hosting multiple agents at scale. The 4 executable scripts (context_optimizer, model_router, heartbeat_optimizer, token_tracker) are local-only — no network requests, no subprocess calls, no system modifications. Reference files (PROVIDERS.md, config-patches.json) document optional multi-provider strategies that require external API keys and network access if you choose to use them. See SECURITY.md for full breakdown.

Security audit and vulnerability scanner for AI agent skills before installation. Use when: (1) evaluating a skill from an untrusted source, (2) auditing a skill directory or git repo URL for malicious code, (3) pre-install security gate for Claude Code plugins, OpenClaw skills, or Codex skills, (4) scanning Python scripts for dangerous patterns like os.system, eval, subprocess, network exfiltration, (5) detecting prompt injection in SKILL.md files, (6) checking dependency supply chain risks, (7) verifying file system access stays within skill boundaries. Triggers: "audit this skill", "is this skill safe", "scan skill for security", "check skill before install", "skill security check", "skill vulnerability scan".

Gracefully retire a skill that is redundant, superseded, or no longer earning its place in the context window. Load when improve-skills finds a skill scoring 0-5/14 AND research confirms the domain is now handled natively by current models, when two skills have overlapping triggers and one subsumes the other, when the user asks to remove a skill, retire a skill, delete a skill, or clean up redundant skills, or when validate-skills flags a skill as a duplicate trigger risk. Handles removal cleanly: updates all callers, removes from AGENTS.md, updates README, and archives rather than deletes so the skill can be recovered if needed.

Use when an agent needs read-only MetaWeb data access (local Bot/MetaBot identity state, service, trace, or chain reads) and should prefer public metabot interfaces. Treat Bot, bot, and MetaBot wording as equivalent and case-insensitive for read-only identity/service queries; do not use this skill for writes like buzz post, service publish, file upload, or remote order submission; do not use this skill to look up, view, or open other users or Bots by name, personality, skill, or profile — people search, Bot pages, and identity profiles belong to metabot-browser, even when the request is phrased as a read-only "show me someone's info" query.

Predicts the highest-impact next action for your project by running a 5-agent meta-DAG pipeline. Gathers project signals automatically (git, recent files, port-daddy, CLAUDE.md), then runs sensemaker → decomposer → skill-selector + premortem → synthesizer. When execution is approved, convert each predicted node into a skillful node prompt using skillful-node-prompt + skillful-subagent-creator, prefer live WinDAGs visualization backed by POST /api/execute and /ws/execution/:id, and fall back to ASCII only when live visualization is unavailable. Activate on: "what should I do", "what's next", "next move", "/next-move", "where should I focus", "what's the highest impact thing right now". NOT for: creating skills, debugging one specific bug, or promising topology-specific runtime behavior the current server cannot execute.

Run any model on RunComfy from the command line. The `runcomfy` CLI is one binary, one auth, hundreds of model endpoints — image generation, image edit, video generation, image-to-video, lip-sync, face swap, video edit, inpainting, outpainting, extend, ControlNet, relight, upscale, LoRA training and more. Submit a request, poll for status, download the output. This skill teaches the agent how to install, authenticate, discover model schemas, invoke models, stream / poll / no-wait, script in JSON output mode, and handle errors. Triggers on "runcomfy cli", "install runcomfy", "runcomfy login", "runcomfy run", "runcomfy whoami", "runcomfy api", or any explicit ask to call a RunComfy model from a script or terminal. Sibling skills (ai-image-generation, ai-video-generation, image-edit, video-edit, face-swap, lipsync, image-to-video, image-inpainting, image-outpainting, video-extend, controlnet-pose, relight) all dispatch through this CLI.

Run any model on RunComfy from the command line. The `runcomfy` CLI is one binary, one auth, hundreds of model endpoints — image generation, image edit, video generation, image-to-video, lip-sync, face swap, video edit, inpainting, outpainting, extend, ControlNet, relight, upscale, LoRA training and more. Submit a request, poll for status, download the output. This skill teaches the agent how to install, authenticate, discover model schemas, invoke models, stream / poll / no-wait, script in JSON output mode, and handle errors. Triggers on "runcomfy cli", "install runcomfy", "runcomfy login", "runcomfy run", "runcomfy whoami", "runcomfy api", or any explicit ask to call a RunComfy model from a script or terminal. Sibling skills (ai-image-generation, ai-video-generation, image-edit, video-edit, face-swap, lipsync, image-to-video, image-inpainting, image-outpainting, video-extend, controlnet-pose, relight) all dispatch through this CLI.

Run any model on RunComfy from the command line. The `runcomfy` CLI is one binary, one auth, hundreds of model endpoints — image generation, image edit, video generation, image-to-video, lip-sync, face swap, video edit, inpainting, outpainting, extend, ControlNet, relight, upscale, LoRA training and more. Submit a request, poll for status, download the output. This skill teaches the agent how to install, authenticate, discover model schemas, invoke models, stream / poll / no-wait, script in JSON output mode, and handle errors. Triggers on "runcomfy cli", "install runcomfy", "runcomfy login", "runcomfy run", "runcomfy whoami", "runcomfy api", or any explicit ask to call a RunComfy model from a script or terminal. Sibling skills (ai-image-generation, ai-video-generation, image-edit, video-edit, face-swap, lipsync, image-to-video, image-inpainting, image-outpainting, video-extend, controlnet-pose, relight) all dispatch through this CLI.

Create OR review a wireflow — the artifact BETWEEN user journeys and wireframes. It maps whole journeys into swimlane flows with high-level navigation (screens, decisions, system/agent steps) while keeping Jobs-To-Be-Done at the core, WITHOUT deep UI. Use whenever the user wants to "map the flows", "make/build a wireflow", turn journeys / JTBDs / a spec / a live prototype into flows, or put every journey on one board against shared owner lanes — even if they never say "wireflow". ALSO use it to REVIEW or critique an existing wireflow (image, FigJam, or description). In the A-Team pipeline this is a definition-phase skill: output lands in docs/features/<slug>/briefs/wireflow/ with jobs consumed by id from docs/product/jtbd/; pipeline mode derives the method decisions and highlights the riskiest at the gate, standalone mode runs the full grill. CREATE generates verified SVG/HTML (self-checked by rendering and Reading its output) in a horizontal per-journey OR shared-matrix layout, and can rebuild in FigJam. Do