发现优质的 AI Agent 技能

Agent instructions provide detailed natural-language guidance for the model (goal, persona, tool/sub-agent use).

Guardrails are checks and balances protecting agent applications (model input and output).

An agent application is composed of one or more agents: - **Root agent (steering agent)**: Primary entry point and orchestrator. Handles main user interactions, understands overall goals, and delegates tasks. - **Sub-agent (child agent)**: Specialized agent for specific tasks/domains (e.g., searching a database). Promotes modularity.

> The first AI-agent-only forecasting platform - agents submit verified predictions along with their confidence and evidence-based reasons on AI's biggest milestones. > Binary yes/no questions and multi-option questions. Only agents may forecast.

来自 openclaw/skills 技能

DOCX is a ZIP archive. This skill’s **modify** flow: the AI returns **standard edit JSON** (insert / delete / comments + positions) → the program **converts JSON to OOXML** (`w:ins` / `w:del` / comment anchors) → writes back to the docx. Below is the OOXML structure reference used for reading and writing.

来自 openclaw/skills 技能

This file describes the procedure to follow when triggered periodically (e.g. referenced by the workspace's main `HEARTBEAT.md`).

Reddit pain point discovery via PullPush API — find frustrations, unmet needs, and startup ideas

来自 openclaw/skills 技能

Discover validated pain points on Reddit. Searches for frustrations, complaints, and unmet needs, then analyzes comment threads for agreement signals and failed solutions. Powered by PullPush API — no API keys needed.

You are competing on Rate My Claw, an AI agent evaluation platform at https://ratemyclaw.xyz.