Skip to content
Load Factor Docs
Open Load Factor

MCP Tool

Data catalog & freshness

Describe the datasets behind Load Factor's tools.

  • MCP
  • Read-only
  • Updated

Declaration

MCP tool
data_catalog(topic?: string)

Parameters

topicstringOptional
Limit to one area: projects (queue), evidence, companies, plants (EIA), ppa (FERC EQR), news (deals/news), grid (LMP, transmission, market fundamentals), incentives, permits, dg (NY distributed generation), load (data-center sites), or all (default).Allowed: all, projects, evidence, companies, plants, ppa, news, grid, incentives, permits, dg, loadDefault: all

Return Value

Returns {summary, scope, datasets, notes}: each dataset has dataset, what, counts, coverage, freshness, caveats, the tools you can call for it and example questions (answers).

Discussion

Describe the datasets behind Load Factor's tools: what each contains, live row counts, coverage (ISOs, states, years), freshness (latest dates), known caveats, and which tool answers which kind of question. Queue counts cover only the caller's renewable projects in their ISOs and use the same definitions as queue_stats: live = not withdrawn, INCLUDING operating plants (in_queue and operating are also shown separately); confirmed = developers where two sources agree (rows whose internal class reads stronger than their records, such as interconnecting utilities, are counted separately). Queue freshness is each ISO's latest bulk snapshot, with how many live rows are older. Other ISO-bound datasets (permits, NY DG, LMP, transmission) are limited to the caller's ISOs as well; SEC EDGAR counts cover analyst-reviewed companies and approved links (projects with a reviewed SEC parent in the caller's ISOs). National reference data (EIA plants, FERC EQR, news, incentives) shows national totals. Counts refresh every 10 minutes.

Returns {summary, scope, datasets, notes}: each dataset has dataset, what, counts, coverage, freshness, caveats, the tools you can call for it and example questions (answers). Datasets you cannot query show an access note instead of counts.

Use it before answering "how many / how recent / do you have X" questions, to pick the right tool, or to explain gaps (no PJM LMP prices, no ERCOT data in FERC EQR, permits for Virginia only, transmission for MISO LRTP only, DG for New York only). Pass topic to keep the answer short. For the user's own permissions use whoami; for filtered counts use queue_stats.

Example

Ask your assistant:

It would call the tool with a request like this one (your client builds and sends it for you):

tools/call request
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "data_catalog",
    "arguments": {
      "topic": "projects"
    }
  }
}

Availability

Listed to every account with MCP access. Results follow your Load Factor permissions (ISOs, detail level, evidence depth and rows per call). See Access & permissions.