fix(reliability): cold-open retry resumes the minted campaign, not a second one (#719)

[100yenadmin]

Closes #719.

The bug (adversarially verified, RRI 2026-06-09)

The shared engine root of two RRI criticals — the hero-bind party-wipe and the input-lock/no-spinner. The DEFAULT cold-open prompt (scripts/play.sh) says start_world("<world>") + "if it returns existing_campaigns, start fresh". When attempt-1 times out (timeout(1) rc=124, >400s) after already minting+seeding a campaign, play.sh's one-shot retry re-runs that identical prompt → the DM mints a SECOND, party-less campaign. The viewer auto-follows the newer empty orphan → "wiped" party + a frozen, input-locked screen.

PR #717 mitigated the read-side (viewer demotes party-less orphans) and #718 added the cold-open spinner — but the orphan was still created (a wasted cold-open + a data-loss footgun). This is the engine-side prevention.

The fix (additive, invariant-safe — engine stays sole writer)

New shared helper clawdnd_coldopen_retry_msg (qa/lib_beat_driver.sh) returns a RESUME directive — get_state(<id>) first, DO NOT start_world, seat a canon PC only if the party is empty (never invent), then open the 2nd-person scene — when and only when this is a DEFAULT cold-open RETRY (first=1, no authored hero) and attempt-1 left a live campaign (clawdnd_live_campaign_id). Otherwise it echoes the base message UNCHANGED:

• first=0 (continuing beat) — never a cold-open re-seed risk;
• HERO_CAMP set — the authored-hero opener already opens on the existing campaign;
• no live campaign — attempt-1 minted nothing, so the normal cold-open must run.

Wired into dm_turn's retry in play.sh. Byte-identical on every path except the one bug. Read-only (asks the engine for the live save) and only on the slow retry path. bash-3.2-clean (printf, no heredoc-in-$() — the documented macOS-system-bash footgun).

The engine seam this relies on already exists + is tested: start_world(resume=) → test_start_world_resume_continues_instead_of_orphaning (test_content.py).

Tests (deterministic, no LLM/network/engine)

5 new shell-helper tests in test_dm_session_remint.py (the pytest-via-/bin/bash pattern):

• resumes the existing campaign on the bug path (campaign_id + DO NOT call start_world + get_state, canon-PC-if-empty; NOT the fresh cold-open prompt);
• UNCHANGED for no-prior-campaign / continuing-beat / authored-hero;
• play.sh wiring anti-drift.

✅ 15/15 in-file · ✅ fast_gate.sh 188 engine + seat-path · ✅ bash -n clean on both scripts.

Moves which RRI gates

zero_critical (the party-wipe + input-lock criticals at their root) → toward no_give_up + cross_persona_sat 4.4→7. Verify on the next VM part-B sweep.

Summary by CodeRabbit

• Bug Fixes

  • Campaign retries now properly resume existing campaigns instead of starting fresh, maintaining party and world state continuity during retry attempts.

• Tests

  • Added comprehensive test coverage for cold-open retry behavior, validating proper campaign resumption and state preservation.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 29ec4ab1-6ce4-4cb1-804b-b41d68ebab7b

📥 Commits

Reviewing files that changed from the base of the PR and between 52e74e2 and a4c5131.

📒 Files selected for processing (3)

• qa/lib_beat_driver.sh
• scripts/play.sh
• servers/engine/tests/test_dm_session_remint.py

---

📝 Walkthrough

Walkthrough

This PR prevents campaign double-minting during cold-open beat retries by detecting existing campaign state and resuming it instead of re-seeding. A new helper function conditions the retry message based on retry type, existing campaigns, and authored hero presence; play.sh's retry path integrates the helper; comprehensive tests validate all conditional paths and wiring.

Changes

Cold-open retry campaign resumption

Layer / File(s)	Summary
Cold-open retry message generator qa/lib_beat_driver.sh	New clawdnd_coldopen_retry_msg() function conditionally outputs a resume directive when retrying a cold-open with an existing campaign and no authored hero, instructing the DM to call get_state, seat a canon hero only if the party is empty, enforce solo-session constraints, and produce opening prose—otherwise returns the base message unchanged.
Retry wiring and validation scripts/play.sh, servers/engine/tests/test_dm_session_remint.py	dm_turn()'s retry block now queries the live campaign id from engine state and rewrites the cold-open retry message via the helper to resume the existing campaign. Five test cases validate the helper's conditional logic (existing campaign, no campaign, continuing beat, authored hero, wiring) to ensure retries correctly resume and abandon the re-seed path.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Possibly related PRs

• electricsheephq/WorldOS#412: Both PRs adjust the cold-open Dungeon Master prompt in scripts/play.sh to ensure solo runs do not auto-recruit a companion, overlapping on shared solo-session instruction logic.

Poem

A rabbit hops through retry lands,
Where campaigns once were double-minted by careless hands.
Now one resume whispers true:
"Keep the state, seat the hero anew—
No orphans left behind, just me and you! 🐰"

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Description check	⚠️ Warning	The description is comprehensive and detailed but is missing the required CLA and licensing checkboxes from the template, though it does address the summary and validation sections.	Add the licensing/CLA checkbox section with the three required statements (CLA agreement, no confidential data, third-party material attribution) even if unchecked.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately describes the main change: fixing a cold-open retry bug where the minted campaign is resumed instead of re-minting a new one, directly addressing issue #719.
Linked Issues check	✅ Passed	The PR successfully implements the fix specified in issue #719: adding clawdnd_coldopen_retry_msg helper to resume existing campaigns on cold-open retries and wiring it into play.sh, with comprehensive deterministic tests covering all paths.
Out of Scope Changes check	✅ Passed	All changes are directly scoped to fixing the cold-open retry double-minting bug: the helper function, its wiring in play.sh, and deterministic tests validating the fix.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}