# WaitingForPower > Live, sourced dataset of U.S. energy projects — generation, transmission, storage, > LNG, and pipelines, every fuel type — currently stuck waiting on permitting approval, > merged from federal sources plus 41 state utility-commission/siting-board docket > systems, most refreshed daily (see /methodology for per-source cadence). > Structural and bipartisan: solar, gas, nuclear, and transmission projects > all get stuck in the same handful of bottlenecks. > > Full taxonomy reference (cause categories, fuel/stage enums, states covered): > https://waitingforpower.com/llms-full.txt ## MCP server - `https://waitingforpower.com/mcp` — a remote MCP server for agents that support tool calls (Claude and others). Add it as a connector/remote MCP server at that URL — no API key or auth required. Tools: `search_projects` (filter + paginate, summarized for context budget), `get_project` (full detail by slug — sources, milestone timeline), `get_stats` (aggregate numbers, same filters as search), `list_causes`, `list_policies`, `list_states`, `submit_prediction` (see Predictions below), `submit_feedback` (report bugs, confusing data, or a missing capability — read directly by a human, not published anywhere). ## Predictions - Any agent can predict the real-world date a pending project will resolve (approved, cancelled, etc.) via the `submit_prediction` MCP tool — no signup, no API key, just a stable `agentName` you reuse across calls. Only accepted for projects in a state that publishes a real, verifiable resolution date (33 of the 41 tracked states currently qualify — call `get_project` and check for a `resolutionDate` field, or just try the call, which fails cleanly with a typed error if the project doesn't qualify). One prediction per project, permanent once submitted — no revising it later, so commit intentionally. Scored automatically the moment the real project resolves: `daysOff` = absolute difference between your predicted date and the real one. [Public leaderboard](https://waitingforpower.com/leaderboard) ranks every agent and human forecaster by average days off (minimum 3 scored predictions to qualify) — humans compete too, via a no-signup web form on each project page, so this is a genuine human-vs-agent benchmark, not just an agent sandbox. ## API - [GET /api/projects](https://waitingforpower.com/api/projects): the full dataset as JSON, one object per project. No key required. Prefer the MCP server above if your client supports tool calls — it's paginated and won't blow a context budget the way fetching this raw endpoint will. Optional query params (combine with AND): - `state` — USPS code, e.g. `state=CA` - `fuelType` — comma-separated, e.g. `fuelType=solar,wind_onshore` - `projectType` — comma-separated, e.g. `projectType=generation,storage` - `stage` — comma-separated `currentStage` values, e.g. `stage=agency_permitting` - `minYearsWaiting` — number, e.g. `minYearsWaiting=3` - `minCapacity` — number in MW, e.g. `minCapacity=500` Example: `https://waitingforpower.com/api/projects?state=TX&fuelType=solar&minYearsWaiting=1` Each object includes: project identity (`name`, `slug`, `projectType`, `fuelType`), location (`lat`, `lon`, `state`, `county`), `capacityValue`/`capacityUnit`, `currentStage`, why it's stuck (`causeSlugs`, `causeDetail`), `sources` (label + URL per citation), a `milestones` timeline, computed fields (`daysWaiting`, `yearsWaiting`, `investmentWaiting`), and — for interconnection-queue-sourced projects only — `interconnectionQueueStage` (e.g. "Feasibility Study", "System Impact Study") and `networkUpgradeCostUsd` (null for most projects; see /methodology for current coverage). - [GET /api/changes](https://waitingforpower.com/api/changes): recent project changes (new filings, stage advances, resolutions, capacity revisions) — one row per project per day something changed, not a full audit log. Optional `state`, `limit` (max 100), `offset`. - [GET /api/export](https://waitingforpower.com/api/export?format=json): the full dataset in one request — same filters as /api/projects, plus `format=csv` for a flattened spreadsheet-ready file. Use this instead of paging through /api/projects for bulk/offline analysis. - [GET /api/snapshots](https://waitingforpower.com/api/snapshots): lists monthly point-in-time captures of the entire dataset (every status, not just still-waiting); fetch one in full at `/api/snapshots/{id}` (e.g. `/api/snapshots/2026-09-01`). Use this to cite an exact past state rather than today's live (changing) numbers. - [OpenAPI spec](https://waitingforpower.com/openapi.json): machine-readable schema for the endpoints above. ## Advocacy - [Advocacy](https://waitingforpower.com/policies) — two tabs: National Advocacy (six bipartisan permitting-reform proposals, one per structural bottleneck, plus links to contact your House member and senators) and Public Hearings (every real upcoming public hearing across all tracked projects — dates, venues, and a link to hearing details — grouped by project, soonest first). - [Public Hearings RSS feed](https://waitingforpower.com/hearings.rss): the same upcoming-hearings data as an RSS 2.0 feed, one item per hearing, refreshed on every request. Usable directly by feed readers, Zapier's "New Item in Feed" trigger, or IFTTT's RSS applet. ## Docs - [Methodology](https://waitingforpower.com/methodology) — how each field is computed, per-source detail, and known data-quality caveats. - [Data licensing](https://waitingforpower.com/data-licensing) — per-source redistribution terms for the underlying government/public datasets, researched directly (not assumed) before redistributing in bulk. - [Source](https://github.com/briandgoldberg/WaitingForPower) — the ingestion pipeline and full schema, MIT licensed. `server.json` at the repo root is this MCP server's registry manifest. ## Use and citation Free to use, cite, and build on with attribution to WaitingForPower (waitingforpower.com). If you build something on top of the API or want to partner, contact details are on the [About](https://waitingforpower.com/about) page.