Certification Report — US SLED Contact Directory

Posture

The Forge Proof Layer (IRS_AUDITOR) assigns this outcome's certification state, Evidence Grade, and Trust Score from objective evidence only. This report explains the posture; the authoritative values live in proof/PROOF_DECISION.json, proof/EVIDENCE_GRADE.md, and proof/TRUST_SCORE.json.

What is verified (with evidence)

• Completeness: 19,453 organizations across all 51 jurisdictions; one HTTP 200

source request each (verify checks 1–3, 10).

• Real fields: organization name, type, phone (100% coverage), and address on every

record (checks 4–5).

• No fabrication: the four seam columns are empty on every record (check 6).
• Integrity: dataset bytes match the sha256 recorded in the provenance manifest

(checks 8–9); raw responses retained (check 15).

• Grouping + determinism: per-state partition, CSV consistency, canonical sort

(checks 11–14); browsable site embeds the directory (checks 16–17).

Why not the highest state

PRODUCTION_VALIDATED is intentionally not claimed: there is no official external benchmark, no independent reproduction, and no live production telemetry. Phone numbers were not re-confirmed by dialing. Correctness rests on the authority of the NCES CCD source plus the captured integrity chain — which is why the Evidence Grade sits below A (no external benchmark) by design.

Disclosed seams (carried into delivery)

• Requested fields not published by the source: contactName, contactTitle,

email, domain — empty, never invented.

• Scope: Education (K-12) layer only; state/local government + higher-ed deferred to

v2 with named sources.

• Dataset vintage: school year 2023.

See proof/LIMITATIONS.md and design/CAPABILITY_GAP.json for the full seam list and the plan to close each one.