MoneyPeel reports MCP server

Read one published report as Markdown: positioning, marketing, growth plays, social, timeline and takeaways. Requires a connected MoneyPeel account. **This call is metered**: each successful read spends one read from the monthly allowance, a single chapter costs the same one read, and reading the same report again spends another. Free accounts get 2 reads per UTC month. Screen with search_reports / list_reports first — they cost nothing — and read only the reports the user actually needs; a failed lookup or a not_found never spends a read. The 'billing.readCharged' field in the result states whether this call spent one (0 means it was a retry inside the idempotency window). It has one authoritative report-content date: reportContentAsOf; reportResearchAsOf and dataAsOf are deprecated aliases carrying the same value, so never present two research dates. Every chapter carries leadAsOf, snapshotType and figuresAsOf: a lead that quotes money, percentages or subscription counts is a historical_research_snapshot, not current revenue. The source block also carries schemaVersion and dataPolicyVersion.

Compare 2-5 companies side by side on MRR, the 30-day amount, business model, economic basis, pricing, acquisition channels, target customer and positioning. Free: no account needed and no reads spent. Every cell carries a status: comparable, partial (e.g. stale, or captured prose rather than a normalised field), not_available (missing stays empty, never 0) or different_basis (the set mixes economic bases, so the amounts must not be ranked against each other). No cell is inferred and no company is ranked.

Remaining reads for the connected account: plan, period, monthly and daily allowance, and the UTC reset time. Free — call it before reading a batch, but do not call it before every read.

Browse the public library, highest revenue first. No account needed, no reads spent, bounded pages. Each row carries the same public summary fields as search. Returns a `nextCursor` while more reports exist. A cursor is bound to the full request, including `limit`: keep the same sort, filters and page size when you page, or the server answers invalid_cursor rather than silently reusing it.

Rank revenue or MRR moves over an explicit window. Free: no account needed and no reads spent. Ranking rule — 7-day board: baseline at least $500 over the previous 7 days, or at least $1,000 over the previous 30 days. 30-day board: baseline at least $1,000 over the previous 30 days. Either way the change must be at least 20%. By default only rows that clear that gate are ranked, and total counts them — the sample is the public growth board (https://moneypeel.com/revenue?by=growth), not the whole catalogue. The default order is percentage — the same order the public growth board prints; absolute (largest USD change) and consistency (steady growth above a one-day spike) are explicit alternatives. Every row carries baseline, start, end, absolute change, percentage, dataPoints, windowStart/windowEnd (null when the figure is source-reported rather than observed daily), growth kind, the economic basis and the metric-level evidence type. Percentages are never returned without their base; stale rows are marked and can be filtered out. MRR has no 7-day series: metric=mrr works with window=30d only. include_ineligible=true returns the wider signal pool instead: rows below the gate are appended after the ranked rows with rank=null, never take a rank, and total becomes the pool size, with counts {eligible, ineligible, pool}. The pool read is capped at 500 rows; truncated=true says the cap was hit.

Search public revenue reports by keyword and/or structured filters. No account needed, no reads spent. Pass "query" for text (name, domain, description) and/or filters such as minMrr or businessModel; at least one of the two is required — to browse the library instead, use list_reports. A filter on a missing value excludes the row (a null MRR is not $0). Each row carries mrr, revenue30d, business model, metric-level evidence type, a revenue snapshot date, revenue freshness, a dataQuality flag, a public preview link and the metered full-report link — often enough to answer without reading a report. When dataQuality.status is "needs_review", quote the caveat before the numbers. When revenueEligibleForRanking is false the amount is stale and must not be cited as current. Structured filters: businessModel, minMrr / maxMrr, minRevenue30d / maxRevenue30d, revenueFreshness, market, economicBasis, basisDetermination, dataQualityStatus, evidenceType, plus sortBy (revenue, mrr, name, growth) and growthWindow. `evidenceType` keeps a row only when every figure it reports (MRR and/or the 30-day amount) carries one of the requested metric-level types; a missing figure is skipped rather than counted as `unknown`. Only the parameters listed in this schema exist — do not invent others. Returns a `nextCursor` when more matches exist; a cursor is bound to the full request — query, filters, sort and `limit` — so keep all of them identical when you page, or it is rejected with invalid_cursor rather than silently reused.