What is a Pancake receipt?

A Pancake receipt is an immutable, shareable backtest artifact at /r/<short_id> that records the strategy spec, evidence dataset hash, full metrics, verification boundary, and a SHA-256 result hash — so any reader can audit or reproduce the run.

When you run a backtest via the Pancake MCP or REST API, the engine emits a receipt: a permanent, URL-addressable record of what was run and what was found. The receipt URL (/r/<short_id>) is stable — it will always return the same result because the underlying row is write-once. In Pancake's hosting model, the receipt is the artifact that travels with the strategy: it validates the backtest and serves as the foundation for advancing a strategy toward live execution (a v2-roadmap capability).

Every receipt contains: the strategy spec (what was executed), the EvidenceDataset reference (rows_sha256 content hash), the full metrics table (Sharpe, Sortino, CAGR, win rate, Wilson CI, Brier score, bootstrap CI, permutation p-value), the verification boundary 3-tuple (verified / agent_supplied_evidence / unmodeled_risks), and the result_hash (SHA-256 of the canonical output envelope).

Receipts are available in three formats: HTML (/r/<short_id>), Markdown (/r/<short_id>.md), and JSON (/api/v1/public/results/<short_id>). Agents can fetch the Markdown format directly; the JSON format is machine-readable for downstream citation tools.

Visibility is controlled per receipt: public receipts appear in the sitemap and llms.txt; unlisted receipts are accessible by URL but not indexed. The cite API at /api/cite/<short_id> produces BibTeX, CSL-JSON, or RIS from any public or unlisted receipt.