MyChart Takeout CAIHL draft report

MyChart Takeout: CAIHL reassessment

September 8, 2026. Published AI-assisted draft.

MyChart Takeout is a deterministic record exporter, so the standalone utility is unscored. Its documented gaps report helps users see missing data. Export supports patient-chosen downstream analysis but does not guarantee a complete institutional record.

Scope and agency posture

Deterministic patient record-export infrastructure, with downstream AI chosen separately. The September 7 closeout places the standalone exporter outside AI scoring scope. Preserve August 14 score 92 only as historical baseline.

Posture: Strongly agency-expanding, with license-clarity, instance-variability, and access caveats.

Axis: Unscored in current comparisons. Historical axis: 92/100. Standalone deterministic exporter. Downstream AI is chosen separately. This is a scope change, not a numerical downgrade.

Sources and documented findings

• The current repository README confirms deterministic export, a local CLI and an in-browser bundle, derived summaries/indexes and no password storage. It describes GAPS.md/gaps.json and diagnostics that report uncaptured endpoints and failures. It identifies instance differences and disabled EHI export on some systems.

• The same README calls the CLI the current primary mode and describes the bookmarklet as an alternative. The repository root does not display a license file. The landing page could not be retrieved, so the hosted build and current simplified flow were not verified. No code was installed or executed.

• The frozen September 7 closeout already recommends that standalone MyChart Takeout remain outside AI scoring scope. This reassessment preserves that methodological decision.

Patient authority

Inference from the documented workflow: Local portable artifacts let the patient choose subsequent analysis and recipients. That is the agency basis, not the maintainer's reputation or an open-source label. Proxy access gives an authorized caregiver capability and does not establish the subject's consent to every later use.

Critical capacity

Editorial assessment: The new documented gaps report matters: a patient can see what was missed rather than assuming silence means absence. Exported originals can support disputes, but extraction does not correct source records or guarantee that an entire institutional history is available.

Informed control

Editorial assessment: The documented local architecture reduces an extra custody layer. It still relies on an authenticated portal and software supply chain, and does not prove the hosted bundle matches source. Optional agents and downstream AI have their own data paths.

Assessment and limits

Apply the September 7 closeout: retain a strongly patient-directed infrastructure description, but remove a comparable AI score for the standalone deterministic exporter. Also correct unconditional complete-record language and credit missing-data visibility. This is a scope correction, not a numerical downgrade.

Confidence: Medium for repository documentation, limited for deployed build.

Remaining uncertainty: Hosted-bundle equivalence, current license grant, proxy semantics and completeness across real portals remain untested.

Published documentation is evidence of stated conditions, not proof of actual implementation. Unknowns did not receive automatic negative points. Funding, sponsorship, and public code do not determine agency by themselves.

Review provenance

• Reviewer/model: OpenAI Codex / GPT-6.
• Method: Focused public-source reassessment using CAIHL: patient authority, critical capacity, and informed control. Existing evidence plus one focused primary-source pass and at most one targeted follow-up. No live product testing. Numerical scores remain provisional editorial placements, not a new calculation.
• Human review: Hugo Campos authorized publication of these AI-assisted draft reassessments on September 8, 2026. This does not claim comprehensive human verification of every finding.
• Earlier review: 2026-08-14. Historical assessment. Earlier claims are not automatically reverified by this publication.