← Về thư mục
📄 / / profiles / r-and-d / raw_evidence / scientific-agent-skills / skills / iso-standards-readiness / references / gap-analysis-checklist.md

Fail-Closed Evidence Review

Research basis: 2026-07-23, extended 2026-07-26 for laboratory lanes. This checklist organizes evidence questions; it is not ISO or IEC text, an audit, an assessment, a legal determination, or a compliance score.

Sections up to "Training, competence, and change control" apply to the device QMS lanes (iso-13485, iso-14971). The laboratory sections near the end apply to iso-17025 and iso-15189. Use the sections your declared profile covers; do not report on a domain you did not sample.

Status vocabulary

Use only:

Never convert these states into a “compliance percentage.” Unequal, conditional, product-specific, and jurisdiction-specific evidence cannot be responsibly reduced to keyword counts.

Required fields for every review item

Field Required content
Item ID Stable unique identifier
Scope Products, sites, processes, suppliers, systems, and period
Owner Accountable role
Status One value from the vocabulary above
Evidence Controlled IDs, revisions/dates, and locations
Source/version Official source, exact edition/version/date, access/currency review
Rationale Evidence-based conclusion or approved not-applicable rationale
Action Gap/change/CAPA ID, owner, due date, and status
Approval Named authorized approver, decision, date, and record ID

Blank, placeholder, inaccessible, stale, or unapproved fields fail closed.

Workflow

  1. Freeze the review purpose and boundaries: internal audit, ISO certification preparation, laboratory accreditation assessment preparation, FDA inspection preparation, national regulatory inspection preparation, MDSAP, or EU conformity assessment. One purpose per review.
  2. Obtain authorized standards and current official regulatory sources.
  3. Record product/site/market/activity scope; route applicability decisions to authorized management, RA/QA, and legal roles.
  4. Build a controlled evidence manifest. Do not classify files from names alone.
  5. Sample implementation records across time, products, sites, shifts, suppliers, and risk as justified.
  6. Trace findings to risk, correction, CAPA, change, validation, training, and management review.
  7. Preserve limitations and unresolved decisions in the report.
  8. Obtain human approval. Do not issue a certification/readiness/compliance claim.

Scope, roles, and source basis

Document and record control

Risk management

Design and development

Supplier and outsourced-process controls

Production, service, validation, and software

Identification and traceability

Complaints, feedback, postmarket, and vigilance

Nonconformity and CAPA

Internal audit and management review

Training, competence, and change control

Current QMSR transition evidence

As of this research date, QMSR is effective and FDA uses Compliance Program 7382.850. Review:

Use scripts/check_qmsr_transition.py; do not create a legacy-QSR clause map as the current control framework.

Laboratory scope, impartiality, and competence (iso-17025, iso-15189)

Metrological traceability, uncertainty, and decision rules (iso-17025, iso-15189)

Methods, validity of results, and reporting (iso-17025, iso-15189)

Medical laboratory pre- and post-examination (iso-15189)

Regime handoff

Before final review, state which authorized party owns the next decision:

Deterministic tooling

Pass --standard iso-13485|iso-14971|iso-17025|iso-15189 to the three profile-aware checks so the domain vocabulary matches the standard under review. An unlisted value is refused rather than defaulted.

All tools are local JSON/Markdown structural checks. Exit 0 means only that no structural finding was generated for the supplied data.