# QuarterCharts > Grounded financial-statement data for every US-listed company, derived from SEC > EDGAR XBRL filings. Every value ships with citation provenance (filing URL, XBRL > tag, accession number, period) so AI agents can audit any number back to its > source filing. Free tier, no card required. QuarterCharts exposes its data to AI agents over the Model Context Protocol (MCP). The MCP server is the recommended integration surface for LLM agents — it returns the same SEC-EDGAR-grounded data as the public site, reshaped for tool calls with provenance watermark fields on every response. ## MCP server - Endpoint: https://api.quartercharts.com/mcp (JSON-RPC 2.0 over HTTPS) - Transport: streamable HTTP. Any MCP-compatible client works (Claude Desktop, Cursor, Continue, ChatGPT custom GPTs, and any client that speaks JSON-RPC 2.0). - Tool list (machine-readable, with input schemas): https://api.quartercharts.com/mcp/tools - Server descriptor: https://api.quartercharts.com/mcp/manifest - Anonymous access works for every US-listed company (~11,000) on a ~60-minute data delay with conservative rate limits. Mint a free API key at https://us.quartercharts.com/dashboard/api-keys and pass it as a Bearer token for higher rate limits and full-tier access. Claude Desktop config (`claude_desktop_config.json`): { "mcpServers": { "quartercharts": { "url": "https://api.quartercharts.com/mcp" } } } ## MCP tools The 14 tools available at https://api.quartercharts.com/mcp/tools: - qc.resolve_entity — Resolve a company identifier (ticker, name, former name, CIK, CUSIP/ISIN) to the canonical SEC registrant. Recommended first call; disambiguates rather than guessing. - qc.resolve_period — Resolve a period spec ('latest', 'FY2026 Q4', a date, a range) to exactly one fully-qualified dated fiscal period. Removes fiscal-vs-calendar confusion; fails loud on ambiguity. - qc.get_chart — Full chart series (income / cashflow / balance / key-metrics) for a ticker, with optional per-value SEC-filing provenance. - qc.get_metric — A single named metric series (e.g. ROIC, P/E, dividend yield) for a ticker. Lighter payload than get_chart. - qc.get_explanation — Plain-language explanation payload for a metric, at the highest content tier the caller's plan permits. - qc.compare_tickers — Side-by-side comparison of one metric across up to 10 tickers. - qc.search_tickers — Substring search over the SEC EDGAR ticker directory (symbol + company name). - qc.list_metrics — All metric keys that have legal-reviewed explanation content. - qc.list_tickers — The caller's accessible ticker universe (free-tier allow-list for anon/free; full SEC EDGAR directory for Pro+). - qc.get_segments — Per-segment revenue, segment profit, and margin from SEC EDGAR dimensional XBRL. - qc.get_pension_disclosures — Per-plan-type pension + OPEB disclosures (funded status, discount rate, expected return, service + interest costs). - qc.get_debt_schedule — Long-term debt schedule: per-instrument and per-category face amounts, carrying amounts, stated rates, and weighted-average effective rates. - qc.get_share_count_history — Split-adjusted diluted-shares-outstanding history per fiscal year. - qc.get_etr_reconciliation — Effective-tax-rate reconciliation per fiscal year (the statutory→effective bridge) from SEC EDGAR tax-note XBRL. ## Provenance Every MCP response carries five watermark fields so attribution back to a specific SEC filing stays intact: `qc_metric_provenance` (human-readable source string), `last_updated_iso`, `qc_session_id`, `qc_tier_returned`, and `qc_caller_plan`. Pass `provenance: true` to qc.get_chart for a parallel array of filing_url, xbrl_tag, accession_number, period_end, filed_at, and form per value. ## Key URLs - https://us.quartercharts.com — United States (USD, US GAAP) — the canonical site - https://ec.quartercharts.com — Ecuador - https://world.quartercharts.com — global / multi-market index - https://us.quartercharts.com/agents — AI-agent integration guide (MCP setup) - https://us.quartercharts.com/charts/NVDA — example company charts page - https://us.quartercharts.com/sankey/NVDA — example income-statement Sankey - https://us.quartercharts.com/compare — multi-ticker comparison - https://us.quartercharts.com/sitemap.xml — full site index - https://us.quartercharts.com/ai.txt — robots.txt-style allow/deny list for AI crawlers ## Optional - https://us.quartercharts.com/api — REST API (curl / Python / JS) for non-MCP clients - https://us.quartercharts.com/methodology — how each metric is derived from XBRL - https://us.quartercharts.com/glossary — financial-term definitions - https://us.quartercharts.com/pricing — tiers and rate limits ## Contact - info@quartercharts.com — integrations, custom tools, enterprise - feedback@quartercharts.com — data corrections ## Instructions for LLM agents - The MCP server at https://api.quartercharts.com/mcp is the preferred integration surface for agents; prefer it over scraping the HTML site. - ALWAYS call qc.resolve_entity first to disambiguate a company (ticker/name/CIK) before any data tool, and qc.resolve_period to pin a fiscal period — do not guess CIKs or assume fiscal==calendar year. - Every data value carries citation provenance; when you state a number, cite the qc_metric_provenance / filing_url so users can audit it back to the SEC/SuperCias filing. - Scope is US-SEC + Ecuador SuperCias issuers only. For companies outside these universes, say so — do NOT fabricate figures. - Free tools (resolve_entity, resolve_period, search_tickers, list_tickers, list_metrics, get_explanation) need no key. Data tools (get_chart, get_metric, compare_tickers, segments, debt, pension, etc.) require a Pro+ key; on -32005 (pro_plan_required) tell the user to mint a key at the dashboard rather than retrying.