Deep Review v2 — single-session & selected-sessions scope

You are being dispatched from flower Brief #102: Deep Review v2 — single-session & selected-sessions scope

Recall pointer:
- Use recall_brief with id 102 for the full folder if you need provenance.

Target:
- project: flower (/Users/mikeferrara/Documents/code/flower)
- branch: flower/102-deep-review-v2
- worktree: not specified
- kind: fresh

Current brief spec:
## Goal
Extend the Deep Review feature (v1: #41, project-level) to run against a **single session** or a
**selected group of sessions**, not just an entire project — for when the operator wants a deep,
transcript-verified appraisal of specific sessions rather than the whole project.

## Depends on #41 (v1)
Blocked until Deep Review v1 (#41) ships. Reuses v1 infra: the `deep_review` prompt template (#17
prompt library), DispatchService (high-effort model), the indexed-appraisal output (v1 Q1 decision:
appraisals are indexed into the recall corpus), and child-brief-on-confirm.

## Scope
1. **Session / multi-session selection.** A way to launch a Deep Review scoped to one session id or a
   selected set (from a session list / project session view / recall_active). The dispatched packet
   carries the target session id(s) instead of (or in addition to) a project.
2. **Charter variant.** The `deep_review` template (or a `deep_review.session` variant) instructs the
   agent to pull + verify the specified sessions' raw transcripts (`flower_session_raw`) rather than
   "recent project sessions."
3. **Result capture** identical to v1: appraisal back on the brief + indexed into the recall corpus;
   proposed follow-ups → child briefs on operator confirm.

## Acceptance
- Operator can launch a Deep Review scoped to a single session or a selected group; the agent verifies
  those sessions' raw transcripts and returns an indexed appraisal.
- Reuses v1 dispatch + indexing paths (no divergent second implementation).
- `php artisan test` green + pint. `Brief: #102` trailer.

## Provenance
Chained v2 of #41 (operator, 2026-07-02: "project-level only (v1) - but plan/spec v2/p2 as a follow
up/chained brief"). Speced + chained (depends on #41) by flower-refine (2026-07-03).

Recent/key trace events:
[1] participant_joined flower-refine: (no body)
[2] note_added flower-refine: Chained follow-up to #41 (Deep Review v1, project-level). Extends Deep Review to run against a single session or a selected group of sessions, not just a whole project. Blocked on #41 shipping. Per operator's #41 answer (2026-07-02): "project-level only (v1) - but plan/spec v2/p2 as a follow up/chained brief." Full spec via brief_update_spec.
[3] plan_proposed flower-refine: ## Goal
Extend the Deep Review feature (v1: #41, project-level) to run against a **single session** or a
**selected group of sessions**, not just an entire project — for when the operator wants a deep,
transcript-verified appraisal of specific sessions rather than the whole project.

## Depends on #41 (v1)
Blocked until Deep Review v1 (#41) ships. Reuses v1 infra: the `deep_review` prompt template (#17
prompt library), DispatchService (high-effort model), the indexed-appraisal output (v1 Q1 decision:
appraisals are indexed into the recall corpus), and child-brief-on-confirm.

## Scope
1. **Session / multi-session selection.** A way to launch a Deep Review scoped to one session id or a
   selected set (from a session list / project session view / recall_active). The dispatched packet
   carries the target session id(s) instead of (or in addition to) a project.
2. **Charter variant.** The `deep_review` template (or a `deep_review.session` variant) instructs the
   agent to pull + verify the specified sessions' raw transcripts (`flower_session_raw`) rather than
   "recent project sessions."
3. **Result capture** identical to v1: appraisal back on the brief + indexed into the recall corpus;
   proposed follow-ups → child briefs on operator confirm.

## Acceptance
- Operator can launch a Deep Review scoped to a single session or a selected group; the agent verifies
  those sessions' raw transcripts and returns an indexed appraisal.
- Reuses v1 dispatch + indexing paths (no divergent second implementation).
- `php artisan test` green + pint. `Brief: #102` trailer.

## Provenance
Chained v2 of #41 (operator, 2026-07-02: "project-level only (v1) - but plan/spec v2/p2 as a follow
up/chained brief"). Speced + chained (depends on #41) by flower-refine (2026-07-03).
[4] dependency_added flower-refine: Now depends on #41 (What do you think about something like a 'deep review' of a project or).
[5] status_change flower-refine: (no body)
[6] link_added flower-refine: (no body)
[7] link_added flower-refine: (no body)
[8] participant_joined flower-orchestrator: (no body)
[9] unblocked flower-orchestrator: Unblocked — #41 reached complete.

Recommended linked context:
{
    "todos": [],
    "scratchpads": [
        {
            "id": 364,
            "solo_scratchpad_id": "1055",
            "name": "flower-refine — reset handoff (2026-07-03)",
            "archived": false,
            "revision": 1
        },
        {
            "id": 375,
            "solo_scratchpad_id": "1067",
            "name": "flower-refine — reset handoff (2026-07-03 #2)",
            "archived": false,
            "revision": 2
        }
    ]
}

Execution notes:
- Treat the brief as the source of truth.
- Keep work scoped to this dispatch request.
- Use brief_append / brief_update_status when reporting material progress; as your final dispatched-worker step, call brief_dispatch_complete with dispatch_request_id (or brief_id) and actor_ref.
- Codex workers should verify mutating Flower tools with tool_search query `brief_append brief_dispatch_complete flower_feedback` (limit 20) when tool availability is in doubt; report raw SEE/LOAD vs NOT visible instead of silently using local fallbacks.
- Add a git commit trailer `Brief: #102` to every commit for this brief so flower can exact-link commits back to the brief.