{"id":"poddubnyoleg-lightdash-mcp","name":"lightdash_mcp","af_score":59.2,"security_score":58.8,"reliability_score":31.2,"what_it_does":"Provides an MCP (Model Context Protocol) server that connects AI assistants to Lightdash to support discovery of projects/explores/schemas, execution of queries, and CRUD-style management of charts, dashboards, tiles, and spaces—primarily via Lightdash REST endpoints using a personal access token, with optional support for auth frontends such as Cloudflare Access and Google Cloud IAP.","best_when":"You already have a Lightdash instance and want an AI agent to programmatically inspect and manage analytics artifacts (charts/dashboards) with an MCP client (e.g., Claude Desktop/Code).","avoid_when":"You cannot safely store and protect LIGHTDASH_TOKEN (or related IAP credentials), or you require explicit documented rate-limit behavior and strict SLAs.","last_evaluated":"2026-03-30T15:35:53.787138+00:00","has_mcp":true,"has_api":false,"auth_methods":["Lightdash Personal Access Token via LIGHTDASH_TOKEN (mapped to HTTP Authorization: ApiKey as described)","Optional Google Cloud IAP JWT signing (IAP_ENABLED + related service-account/ADC env vars) with Proxy-Authorization: Bearer <jwt>","Optional Cloudflare Access (CF_ACCESS_CLIENT_ID / CF_ACCESS_CLIENT_SECRET)"],"has_free_tier":false,"known_gotchas":["State-changing tool calls (create/update/delete) are likely not idempotent; agents should avoid automatic retries without safeguards.","Credentials in MCP client config files (e.g., .mcp.json) risk leakage if not gitignored.","The server requires LIGHTDASH_TOKEN and LIGHTDASH_URL; incorrect/expired tokens will lead to 401 errors.","If your Lightdash instance is behind Cloudflare Access or Google IAP, you must set the corresponding environment variables or connectivity/auth will fail."],"error_quality":0.0}