fix(kanban): --takeover re-homes the card; explicit --session beats parent home (t_5c908e14)

[Kyzcreig]

fix(kanban): --takeover re-homes the card; explicit --session beats parent home (t_5c908e14)

A --takeover on assign/unblock/promote/reclaim/triage-resolve/complete now
sets tasks.session_id to the taking session, records prev_session_id and
session_id in the takeover event payload, notes the re-home in the audit
comment, and subscribes the taker's chat (best effort). Other verbs stay
one-off foreign actions; --operator never re-homes; edit --session <sid>
remains the status-free re-home.

Children created with --parent already inherit the parent's CURRENT home,
so after a takeover they follow the new home. create --session <sid>
(explicit) now wins over the parent's home; an omitted --session still
defers to the parent. create_task gains session_explicit=False (default
keeps library/tool behavior unchanged).

The origin: line is already written by create_task (the single create
choke point) when the body lacks one; a CLI test now pins that.

Verified: tests/hermes_cli/test_kanban_home_session.py 89 passed.
Mutant (new_home = None, i.e. skip the re-home): 7 tests RED.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[Kyzcreig]

CHANGES REQUESTED (Apollo diff review 14:5x PT, PR #1118 @eb246a2b). Core is right (explicit create --session beats parent; children follow the CURRENT home; prev_session_id in the takeover event). Two defects, both from the card's own 'sweeps never re-home' line:

1. CRON / SWEEP actors must NOT re-home. scripts/mq-review-card-closer.sh (cron 1cd274c546db) runs 'complete --takeover' on review cards every 20 min; as written every card it closes is re-homed to a cron_* session, pulled out of its home conversation's count, and its pings go nowhere. Skip the re-home when the actor session id starts with 'cron_' or the caller is a delegate child/sweep (the takeover event still records the actor). Test: cron-session actor -> session_id unchanged.
2. Drop 'complete' from REHOME_ON_TAKEOVER_ACTIONS. Completing is terminal — re-homing a finished card only moves it between conversations' counts. Adoption of an abandoned card happens on assign/unblock/promote/reclaim/triage-resolve (or edit --session). Test: complete --takeover -> session_id unchanged.
   Keep the rest. Push to the same branch; CI is the gate; hand back.

[ang-fleet-lander]

🤖 merged-by: apollo · lane: kanban-merge-pass · gate: BYPASS: FleetReview paused since 09-22; gate = CI green + Apollo diff review · why: t_5c908e14: --takeover re-homes (not on complete; cron/delegate never adopt); Apollo reviewed; rebased onto flake fixes

[ang-prism]

FleetReview

Review: post-merge · head 1bfad2aa34f2 · duration 13m 33s
Profile: light (merit: default light: lines 262<800, files 3<1000000, hunks 12<1000000, no hot path) · policy: below-size-and-path-gates
Roster: B-assert-ctx → gpt-6-sol (openai), C-assert-xhigh → claude-code-opus-5-5 (anthropic), G → grok-4.6 (xai), L6 → gpt-6-sol (openai)

Post-merge review (fleetreview:post-merge override): this reviewed the merge commit against its first parent — the bytes that already shipped. It is not a pre-merge gate pass.

profile: light (rule: default light: lines 262<800, files 3<1000000, hunks 12<1000000, no hot path) · round 0 · members: B-assert-ctx, L6, C-assert-xhigh, G · families: anthropic,openai,xai

Confidence: 3/5

Findings

• P1 hermes_cli/kanban_db.py:5088 — Unstamped child · agreed: B-assert-ctx,C-assert-xhigh,G (openai, anthropic, xai)
• P1 hermes_cli/kanban_db.py:5730 — Unhomed adoption · agreed: B-assert-ctx,C-assert-xhigh (openai, anthropic)

---

FleetReview provenance · models: B=gpt-6-sol, C=claude-code-opus-5-5, D=grok-4.6 · cost: $1.19 · duration: 13m 28s · rounds: 1 · files examined: 3