planning/interview: aihero course lane 5: grilling parity check

[kyle-sexton]

Context

Fifth 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 grilling family is the attributed behavioral source of planning:interview (frontier rounds, facts-vs-decisions split, confirmation gate; SSOT rows 3-4), under our no-grill vocabulary constraint. The 2026-08-17 upstream recheck confirmed the mapping intact and found only cosmetic upstream drift since v1.2.3. Expected outcome: mostly confirmation; small tree.

Proposed work

• Diff the course lesson's grill-execute-clear narrative (design tree, frontier rounds, empty-frontier stop, execute, clear) against planning:interview and the workflow's spec-first mode; confirm each element present or deliberately different
• Check course-only elements for adoption value (the grill-execute-clear loop framing itself; anything the lesson states that his repo skills do not)
• Confirm the emoji-anchors opt-in and answer-by-number affordances still match his current shape (per the recheck, his changes since audit are cosmetic)

Acceptance criteria

• Adopt / reject-with-reason / track-on-event rows recorded in docs/upstream/aihero-course.md for every grilling-lesson claim
• Work items filed for any decided plugin change (none decided; recorded explicitly)
• Lane summary written to the lane's topic slice

References

• docs/topics/pocock-course-lanes/PLAN.md (contract + recheck record)
• plugins/planning/skills/interview/SKILL.md and docs/upstream/mattpocock-skills.md rows 3-4 (grilling-family provenance)
• Upstream: mattpocock/skills skills/productivity/grilling/SKILL.md, grill-me/SKILL.md

Metadata

Field	Value
Category	unspecified
Area	planning
Ecosystem	unspecified

[kyle-sexton]

Lane 5 closed (2026-08-18, core-lane chain on claude/pocock-steering-course-00zkvd). Interview-first per contract, plus a user-invoked /planning:audit-answers adversarial pass — two fresh-context validators over the whole answer set with rationale withheld, both fetching the upstream grilling-family primaries independently.

The audit earned its keep: the expected "mostly confirmation" held for ~20 techniques (parity / stronger / deliberately-different, all evidence-cited), but the provisional zero-work-items answer was OVERTURNED — the mid-grilling prototype detour and plan-mode-off guidance are genuinely absent from planning:interview, filed as #2998 (includes reconciling planning:plan's clarifying-rounds-in-plan-mode line). Two evidence corrections persisted (confirmation gate is parity, not a house strengthener; the course's "clear" step is unverified and upstream's own docs teach same-conversation-into-spec — loop-position nuance recorded). Adjacent catch: the SSOT's PR-#532 "house additions" provenance claim is contradicted on current upstream main — annotated.

Acceptance criteria: ☑ decision rows recorded (docs/upstream/aihero-core-lanes.md, commit 069b7f9d; lane 6 harvests) · ☑ work items filed (#2998) · ☑ lane summary (this comment + record section). Tracked residue: the course lesson itself is egress-blocked from the vetting environment — the diff ran against upstream repo primaries; the "clear"/course-only row re-opens if the lesson text is supplied.

---

Generated by Claude Code