spicy-docs

Collect government-source records, preserve their fields and evidence, and publish immutable releases that can be checked offline. Use the records directly or supply them to a dataset application such as DocSpec.

Try it offline

Use Python 3.12 and uv. From this checkout:

uv sync --frozen --all-extras
uv run --frozen python examples/offline_release.py
./scripts/check

The example publishes, verifies and reads one synthetic GAO page. It needs no credentials or network access. The JSON output contains the source record, literal publisher topic, evidence references and retained file paths.

Try --case unexpected to preserve a different topic, or --case missing to inspect an expected refusal and its retained bytes. Use --directory /path/to/example to choose an output parent; its gao/ child must be new. The example's fixed implementation ID is for demonstration only.

Choose your next step

• Make a change: Contributing.
• Find the code: Architecture.
• Choose an output: Source workflows.
• Run a command: Operator guide.
• Install a smaller reader: Installation.
• Find a source rule: Documentation.

Sources and supply

Source releases support Federal Register, Regulations.gov through Mirrulations, captured spicy-regs public comments, named GAO product pages, and a retained OpenFEC committee census. Raw Mirrulations, CourtListener and FEC readers are also available.

The bill API separately captures BILLSTATUS XML and a selected bill-text version, preserving identity, source fields and bytes. The CFR/eCFR API captures explicitly selected regulation XML, retaining printed dates separately from requested dates and editions. Its GovInfo metadata mapping supplies package and constituent records, including repeated fields, links and publisher notes.

Prefer community spicy-regs tables where they carry the required data; use origin acquisition for missing coverage. Capture and pin whole named agency partitions before classifying rows. Source choice is explicit: the CLI does not perform automatic fallback. See supply policy.

Live GAO capture needs ZYTE_TOKEN. Keep credentials out of locators, artifacts, logs and errors; follow the acquisition rules.

Provenance

Extracted from spicy-regs branch integrate/payload-prereqs, commit ff8d202 and neighboring changes. Earlier history remains there. Some format identifiers retain the spicy-regs name; current releases require producer spicy-docs and current formats.