idkmesh

IDKGraph P1 independent-review follow-up

Date: 2026-08-28 Repository: MSKazemi/idkmesh

User direction

Continue improving IDKMesh from the current repository state.

Repository state observed

• P0 IDKGraph Issue #20 is complete.
• PR #130 was closed unmerged because it duplicated the canonical identity layer; later T1–T5 work landed separately.
• Residual deterministic health checks landed through PR #149.
• Accepted-ADR linkage debt was reviewed and corrected through PR #155.
• A deterministic 15-item orphan-warning cohort was frozen through PR #162.
• That cohort was classified/corrected through PR #166:
  • 6 confirmed navigation gaps;
  • 7 intentional project-memory records;
  • 2 reference-evidence records;
  • 0 uncertain.
• Issue #152 remains open because automated/AI-assisted review cannot truthfully invent independent human reviewer time or independent agreement.
• Issue #35 remains an external GitHub administration gate for branch protection/rulesets before stronger autonomous writes.

Action taken

Created a bounded independent-review experiment instead of adding more automatic repair:

• docs/research/IDKGRAPH_P1_INDEPENDENT_REVIEW_PROTOCOL.md
• examples/idkgraph-p1-review-session.example.json

The protocol freezes the same 15-item cohort, defines reviewer independence and anchoring disclosures, requires self-reported active review minutes rather than inferring attention from GitHub timestamps, and specifies descriptive agreement/action-cost metrics.

Scientific boundary

The objective is not to maximize agreement with the original PR #166 classification. Disagreement, uncertainty, and higher review cost are valid evidence.

The first milestone is one complete eligible independent 15-item review. Further reviews should be retained rather than selectively discarded.

Safety boundary

No automatic classification, link insertion, move, deletion, issue closure, or merge authority is added. The experiment collects review evidence only.

This site is open source. Improve this page.