{"id":"pinkpixel-dev-taskflow-mcp","name":"taskflow-mcp","af_score":59.5,"security_score":24.8,"reliability_score":36.2,"what_it_does":"TaskFlow MCP is a Model Context Protocol (MCP) server that helps an AI assistant plan and execute task workflows. It creates requests with tasks/subtasks (including dependencies and notes), persists them to disk (JSON/YAML), enforces an approval-centric workflow, supports exporting status reports, and includes an archive system for completed requests.","best_when":"You want a local or per-project MCP tool that persistently manages task plans and guides an assistant through a structured task/approval loop.","avoid_when":"You need strong server-side access control, multi-tenant isolation, or guaranteed concurrency safety across many simultaneous users/processes.","last_evaluated":"2026-03-30T15:28:59.327508+00:00","has_mcp":true,"has_api":false,"auth_methods":["None explicitly described (local MCP server started via CLI)"],"has_free_tier":false,"known_gotchas":["Because persistence is file-based (tasks.yaml/tasks-archive.*), multiple concurrent MCP clients pointing at the same file/path may overwrite each other.","Approval-oriented workflow implies the assistant should pause for user confirmation; if the client ignores this, tasks may be marked done prematurely.","Export operations write to an outputPath; ensure outputPath permissions/paths are correct to avoid failures.","Delete/update operations may permanently remove data from the active request; be sure to understand when archiving is required."],"error_quality":0.0}