Require provenance for spreadsheet derived estimates

[EdbertChan]

Summary

Each number in the sheet now has to say where it came from. A filled in average must name how it was made. A guess cannot stand in for a real source. The check fails when a needed name is missing.

Review Claim

The spreadsheet schema and validator now require provenance for derived estimates and keep them out of source-coverage claims.

Review Lane

behavior

Review Unit

product-skill

Safety Invariant

No derived value can appear without an explicit method, lineage, and derivation note; derived values cannot satisfy source-backed competitor coverage.

Slice Rationale

This PR contains the schema and validator behavior. The separate engine-runtime PR contains the completion guard.

Non-goals

• No live Google Sheet writes.
• No change to published source values.
• No inference of missing competitor observations.

Test Plan

Test Plan

• python3 -m unittest product/skills/spreadsheet-authoring/tests/test_validate_schema.py
• python3 -m py_compile product/skills/spreadsheet-authoring/scripts/validate_schema.py

Revert Plan

Revert Plan

• Safe to revert? Yes.
• Revert command: git revert
• Post-revert steps: None.
• Data migration? No.

---

Note

Low Risk
Documentation and offline CSV validation for a product skill; no runtime engine, auth, or live sheet writes in this change.

Overview
Introduces the spreadsheet-authoring product skill: a workbook modeling guide that treats sheets as raw observations → aggregation table → chart, with typed Period Start/Period End, separate Observation Point qualifiers, and strict rules for derived estimates (midpoint / linear interpolation only, with method, lineage, and notes).

Adds validate_schema.py for CSV/fixture preflight: required columns, no qualifiers in date columns, unique observation keys, source URLs on numeric rows, derived-row provenance checks, and optional --expected-entities coverage where derived estimates do not count as source-backed competitor coverage. Failed coverage is meant to surface as SPREADSHEET_COVERAGE: FAIL rather than zero-fill or infer missing entities.

Includes skill invoke/skip test notes and unit tests for the validator behavior.

^{Reviewed by Cursor Bugbot for commit c13bc11. Bugbot is set up for automated code reviews on this repo. Configure here.}

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: serverGenReqId_3cb91070-39cc-45a8-97a9-924dde72a1dc)

[mergify]

Queued — the merge queue status continues in this comment ↓.

[EdbertChan]

@Mergifyio queue

[mergify]

Merge Queue Status

• ✅ Entered queue — 2026-09-29 04:39 UTC · Rule: admin-bypass · triggered by @EdbertChan with the @mergifyio queue command
• ✅ Checks passed · on draft merge queue: checking #1219 on main (b0d44d5) #1223
• ✅ Merged — 2026-09-29 05:23 UTC · at 1c55fa547966d2cc96aa9754c27603755facb8f0 · squash

This pull request spent 44 minutes 29 seconds in the queue, including 44 minutes 3 seconds running CI.

Required conditions to merge

• check-success = lint
• check-success = test
• check-success = validate