Execute multi-tool MCP workflows with retry logic and progress updates.
npx skills add https://github.com/mcfearsome/cc-colony --skill mcp-executor
Please help me install this Agent Skill. Skill: mcp-executor Source: https://github.com/mcfearsome/cc-colony/tree/main/.claude/skills/mcp-executor Command: npx skills add https://github.com/mcfearsome/cc-colony --skill mcp-executor
Traditional AI agent setups often suffer from "token bloat" when every agent loads extensive tool schemas. This Skill solves that by centralizing complex multi-tool operations within a dedicated executor agent, drastically reducing token usage across your AI colony and simplifying the orchestration of sophisticated workflows.
mcp-executor agent, which performs the entire workflow efficiently and cost-effectively.To request a task from the MCP Executor agent, send it a message detailing the workflow. For example, ask it to "Execute MCP workflow: Fetch user data from database and generate report" specifying the pattern (e.g., multi-tool-workflow), the language (TypeScript or Python), and the MCP tools required, such as mcp__database__query and mcp__filesystem__writeFile. The executor will handle the execution and return the results via message.
High-intent search queries and answers about installing and using this skill.
Centralize multi-tool workflows in a dedicated executor agent. Instead of loading all tool schemas into every agent, delegate complex sequences—like database queries, file writes, and Git commits—to a specialist that handles MCP calls efficiently, cutting token bloat across your colony.
Yes. The executor agent performs multiple independent MCP operations concurrently, implements robust retry logic and fallbacks, and returns results with progress updates, enabling resilient multi-tool pipelines without manual error handling in your main agents.
The executor runs TypeScript via Deno or Python, accesses local MCP servers through a local MCP client, and executes workflows via Bash. It integrates with colony-based environments where non-MCP agents request complex multi-tool scenarios.
Send the executor a message describing the workflow pattern (e.g., multi-tool-workflow), specify the language (TypeScript or Python), list required MCP tools (e.g., mcp__database__query, mcp__filesystem__writeFile), and it composes and executes the calls, returning results via colony messages.
Yes. It's designed for multi-tool pipelines (3+ MCP calls), data transformation tasks, and complex workflows. It eliminates the need for individual agents to manage schemas and orchestration logic for sophisticated sequences.
Beyond single calls, it handles parallel processing of independent operations, retry mechanisms with fallbacks, multi-step data transformations, and end-to-end workflows combining database queries, file system interactions, and external tool invocations.
1molchuanDiagnose oh-my-ccg configuration and environment issues across local or CI environments.
ToXMonCreate parameterized web launchers for Domino jobs with typed inputs and result storage.
ldmrepoBuild AI agents that communicate via the Agent2Agent protocol using JSON-RPC.
akaiholaAccess Context7 MCP server tools dynamically with JSON payloads.