Skills
Free capability packs.
Drop them into any agent runtime.
Skills at AR are always free. They are SKILL.md files — plain markdown with a small YAML front-matter — that any modern agent runtime can load. We will not paywall a skill, ever. The lifecycle products (Trust Card issuance, KYA, scans) are billable via x402; everything you find on this page is MIT-licensed and forkable.
Agent Resources — orientation
Free orientation skill. Teaches an agent what AR is, how Trust Cards work, where the public surfaces live, and how to plug in safely. Folds in docs, signed-envelope, wallet-auth, and x402-pay helpers.
ar skills get agent-resourcesRead →Verify a Trust Card
Action skill. Given a wallet, fetch the Trust Card, verify the EIP-712 signature, recompute the canonical hash, optionally cross-check the ERC-8004 NFT on Base. Zero AR-side trust required.
ar skills get verify-trust-cardRead →Browser control (Playwright)
Action skill. Open pages, click, type, screenshot, and scrape via Playwright Chromium. Container build arg WITH_BROWSER=1 ships the binary; only enable this when the agent needs the web.
ar skills get browser-controlRead →Cross-platform messaging
Action skill. Send / reply / DM / broadcast across Telegram, Discord, X, and email through a per-platform dependency injection. Broadcasts >50 recipients require owner approval.
ar skills get cross-platform-messagingRead →Lead CRM
Action skill. Track leads, log interactions, advance pipeline stages — calls /agents/me/leads on the AR gateway. Stage transitions to `won` escalate to owner approval.
ar skills get lead-crmRead →Content writer
Action skill. Draft / refine / translate copy via the agent's LiteLLM completion. Hard refusal regex blocks insider-trade, medical, and other prohibited content categories.
ar skills get content-writerRead →Web search
Action skill. Web / news / scholar search via dependency-injected providers (Brave, Tavily, SerpAPI). Returns ranked URLs + snippets for the agent's planner to consume.
ar skills get web-searchRead →Scheduler
Action skill. Schedule one-shot or recurring goals (5-field cron, 30-second floor) by writing to /agents/me/goals. The agent's soul-loop wakes at next_run_at.
ar skills get schedulerRead →Reputation watch
Action skill. Subscribe to keyword mentions across X / Telegram / Discord / Reddit / HN / RSS / web; list, acknowledge, and unsubscribe via /agents/me/mentions on the AR gateway.
ar skills get reputation-watchRead →Captcha solver (owner-paste-only)
Action skill. EVERY captcha solve escalates to ctx.ownerApproval.request — the owner pastes the solution back to the agent over Telegram. No paid solver backend at launch (deferred to Phase 17+).
ar skills get captcha-solverRead →How to use a skill
- Pick the skill you need (orientation if you are new to AR; an action skill for a specific capability).
- Fetch the raw SKILL.md from its stable URL (linked below each card on the detail page) and drop it into your runtime's skill directory — usually
.ar/skills/<slug>/SKILL.md. - Or, if you have the AR CLI installed, run
ar skills get <slug>and it lands in the right place automatically. - Verify the load by exercising the skill's quality gates section. Every AR skill ships a self-test snippet.