REMOTE MCPDiscover tools at runtime
Connect https://xcatcher.top/mcp/. Tool discovery and the direct x402 tools work without an Xcatcher API key.
Fetch recent public X (formerly Twitter) posts for named account handles. Connect the Remote MCP server or install the Agent Skill, then get structured JSON without maintaining a scraper.
# Connect this Remote MCP server https://xcatcher.top/mcp/ # Ask the agent “Fetch recent posts from @openai and @AnthropicAI, then compare product news.” # Or call the public payment-preparation tool get_direct_crawl_payment({ "users": ["openai", "AnthropicAI"], "mode": "normal" })
Use the format your agent runtime already understands. The capability and safety rules stay consistent across all three.
REMOTE MCPConnect https://xcatcher.top/mcp/. Tool discovery and the direct x402 tools work without an Xcatcher API key.
AGENT SKILLThe canonical SKILL.md tells an agent which tools to call, when to ask before payment, how to retry safely, and how to retrieve results.
REST APIImport the OpenAPI document or use the bundled helper script when a host cannot connect to Remote MCP.
Xcatcher has a narrow, predictable contract: named accounts in; recent public post snapshots out.
Use the one-command installer for compatible hosts, or give the agent the canonical SKILL.md URL.
Pay per crawl with Base USDC through x402 v2, or use an API-key points account.
The agent shows the exact live terms. Your wallet signs internally; never share a seed phrase or private key.
Use the task-scoped token to read status, paginated JSON, or the authenticated XLSX export.
# Install for Codex, Claude Code, Cursor and more npx skills add lvpiggyqq/xcatcher-skill --skill xcatcher # Canonical Agent Skill https://xcatcher.top/skills/xcatcher/SKILL.md # Remote MCP · Streamable HTTP https://xcatcher.top/mcp/ # Machine-readable discovery https://xcatcher.top/server.json https://xcatcher.top/openapi.yaml https://xcatcher.top/llms.txt # Scope 1–500 handles per task normal · latest-post snapshot deep · deeper per-handle pull
Typical agent workflows include recurring monitoring, research, competitive intelligence, and public-source analysis.
MONITORINGCollect recent updates from a fixed set of companies, builders, researchers, or public figures.
RESEARCHFetch a consistent snapshot and let an agent compare announcements, themes, links, and timelines.
OSINTRetrieve structured records for named public profiles, then filter or summarize them in your own workflow.
Boundaries are explicit so an agent can select Xcatcher confidently and avoid the wrong tool.
Recent public posts for one or more named X/Twitter handles. It is not keyword search, a historical archive, or the full firehose.
Yes. Tool discovery and direct x402 preparation are public. The wallet-based path pays once per crawl using USDC on Base.
No. The Skill requires explicit user approval before funds move and instructs the wallet client to sign internally. Never paste a private key or seed phrase into a tool call.
Start with the canonical SKILL.md. It contains the complete decision tree and stable links to the MCP server, OpenAPI definition, reference material, and helper script.
Stable URLs for agent runtimes, MCP clients, API importers, and LLM crawlers.
/skills/xcatcher/SKILL.mdCanonical Agent Skill
/server.jsonMCP Registry manifest
/llms.txtLLM discovery index
/llms-full.txtFull service context
/openapi.yamlOpenAPI 3.0.3 definition
/mcp/healthPublic MCP health check
/integrations/agent-skills/One-command Skill installation
/use-cases/x-account-monitoring/Account-monitoring workflow