planning: aihero course lane 4: plan-mode asset-rush critique vs interview-first sequencing

[kyle-sexton]

Context

Fourth of six vetting lanes for the AI Hero course lessons; contract in docs/topics/pocock-course-lanes/PLAN.md (branch claude/plan-mode-discussion-55kszx). His "Why Plan Mode Sucks" lesson argues plan mode rushes to an asset (the plan reads like the implementation) with no alignment step, driven by the sycophantic asset-rush trait; his design-concept framing (shared understanding, not an artifact) is the alternative. Our pipeline already sequences planning:interview (pre-clarity contract) upstream of planning:plan.

Proposed work

• Vet the critique against our sequencing: does the interview-first pipeline fully answer the asset-rush failure, and where does plan mode (the harness feature) still fit for us
• Audit whether lock mode and auto-detect's synthesize-directly path are licensed exceptions or quiet reintroductions of the rush (the auto-guard and STOP-on-gap rules are the existing defenses; check they hold)
• Decide whether the design-concept idea earns anything beyond the general-domain interview's shared-understanding terminal (feeds lane 6 vocabulary)
• Verified plan-mode behavior claims arrive from the pre-dispatched harness-claims research

Acceptance criteria

• Adopt / reject-with-reason / track-on-event rows recorded in docs/upstream/aihero-course.md for every plan-mode-lesson claim
• Work items filed for each decided plugin change (none decided; recorded explicitly)
• Lane summary written to the lane's topic slice
• The lock/auto-synthesize audit verdict is recorded (licensed exception vs gap, with evidence)

References

• docs/topics/pocock-course-lanes/PLAN.md (contract + lane definition)
• plugins/planning/skills/interview/SKILL.md (pre-clarity stance, auto-guard, lock mode)
• Harness-claims research artifact: .work/pocock-course-lanes/harness-claims/RESEARCH.md (memory tier)

Metadata

Field	Value
Category	unspecified
Area	planning
Ecosystem	unspecified

[kyle-sexton]

Lane 4 closed (2026-08-17, core-lane chain on claude/pocock-steering-course-00zkvd). Interview-first per contract: 1 round, 5 questions, register gate clean, user confirmed.

Verdicts (full rows in docs/upstream/aihero-core-lanes.md, commit c18d7856 — lane 6 harvests): asset-rush critique ADOPTED as convergent-validated — the interview-first pipeline is the alignment step his lesson says plan mode lacks; plan mode RETAINED as permission enforcement + approval UI only (already codified in planning:plan); the lock/auto-synthesize audit lands LICENSED EXCEPTION on four cited structural defenses, with regression coverage filed as #2997; design-concept framing earns no new machinery, term routed to lane 6 (#2904); C7/C8 CONFIRMED, C9 PARTIALLY TRUE with the no-plan-view correction.

Acceptance criteria: ☑ decision rows recorded (core-lane record, pending lane 6's collection) · ☑ work items filed for each decided change (#2997) · ☑ lane summary (this comment + the record section) · ☑ lock/auto-synthesize audit verdict recorded with evidence.

---

Generated by Claude Code