Bump Codex package to 0.4.4

Author: Cipher-85

.agents/skills/playtest-report/SKILL.md

@@ -36,7 +36,14 @@ Generate this template and output it to the user:
 - **Session Type**: [First time / Returning / Targeted test]
 
 ## Test Focus
-[What specific features or flows were being tested]
+- **Hypothesis**: [What feeling, behavior, or evidence this playtest is probing]
+- **Setup/build**: [Build, commit, command, save state, or scenario if known]
+- **Observation prompts**:
+  1. [Specific thing to watch for]
+  2. [Specific thing to watch for]
+  3. [Optional specific thing to watch for]
+  4. [Optional specific thing to watch for]
+- **Verdict/evidence to return**: [User-owned pass/fail/needs-rethink verdict plus notes, screenshots, logs, or report path]
 
 ## First Impressions (First 5 minutes)
 - **Understood the goal?** [Yes/No/Partially]
@@ -91,6 +98,12 @@ Generate this template and output it to the user:
 
 Read the raw notes at the provided path. Cross-reference with existing design documents. Fill in the template above with structured findings. Flag any playtest observations that conflict with design intent.
 
+If the notes omit a specific hypothesis or focus, state that gap before
+analysis and ask for or infer only a provisional `Playtest focus:` from the
+notes. Do not route the user to another manual playtest without a concrete
+hypothesis, setup/build if known, 2-4 observation prompts, and the
+verdict/evidence the user should return.
+
 ---
 
 ## Phase 3: Action Routing
@@ -141,6 +154,10 @@ Verdict: **COMPLETE** — playtest report generated.
 - Act on the highest-priority finding category first.
 - After addressing design changes: re-run `$design-review` on the updated GDD.
 - After fixing bugs: re-run `$bug-triage` to update priorities.
+- If another user-owned playtest is the routed next step, include `Playtest
+  focus:` with the specific hypothesis, setup/build if known, 2-4 observation
+  prompts, and the verdict/evidence the user should return. The focus brief
+  narrows the test but leaves game-feel and balance decisions to the user.
 
 ## Ported metadata
 

.codex/VERSION

@@ -1 +1 @@
-0.4.3
+0.4.4

.codex/docs/session-continuity.md

@@ -26,6 +26,24 @@ The checkpoint exception is narrow. It never authorizes new design, game-feel,
 balance, architecture, source, registry, index, status-file, commit, push,
 branch, build, boot-smoke, mutating `gh`, or additional file changes.
 
+## User-Owned Playtest Focus
+
+When owed verification or the next valid lane is a user-owned playtest, preserve
+a concrete focus brief in both the closeout and any `## Session Worklist` entry.
+Use the label `Playtest focus:` and include:
+
+- **Hypothesis**: what feeling, behavior, or evidence the playtest is probing.
+- **Setup/build**: the build, command, save state, or scenario to use when
+  known.
+- **Observation prompts**: 2-4 observation prompts for specific things the
+  user should watch for.
+- **Verdict/evidence to return**: the user-owned pass/fail/needs-rethink
+  verdict plus the notes, screenshots, logs, or playtest report path needed to
+  make the evidence usable.
+
+The brief narrows the test; it does not make the game-feel, balance, keep,
+revert, or tune decision for the user.
+
 ## Pause Procedure
 
 Before pausing a meaningful work unit, check whether the invoked workflow still
@@ -44,6 +62,8 @@ decision, blocker, or true stop point.
    clear next lane:
    `Next action:` then `1. (Recommended) [action label] - [brief reason /
    command]`. The user can reply with `1`.
+   If that lane is a user-owned playtest, include the preserved `Playtest
+   focus:` brief before the next-action prompt.
 4. Preserve exact next commands only when they are known to be useful.
 5. Keep local-only notes out of tracked docs unless they are project state.
 6. Suggest `$handoff [short-label]` when installed and the next session would

.codex/lib/validate_runtime.py

@@ -227,6 +227,34 @@
     "numbered choice set",
 )
 
+PLAYTEST_FOCUS_CONTRACT = {
+    "AGENTS.md": (
+        "user-owned playtest",
+        "Playtest focus:",
+        "hypothesis",
+        "setup/build",
+        "2-4 observation",
+        "verdict/evidence",
+    ),
+    ".codex/docs/session-continuity.md": (
+        "user-owned playtest",
+        "Playtest focus:",
+        "hypothesis",
+        "setup/build",
+        "2-4 observation",
+        "verdict/evidence",
+        "Session Worklist",
+    ),
+    ".agents/skills/playtest-report/SKILL.md": (
+        "user-owned playtest",
+        "Playtest focus:",
+        "hypothesis",
+        "setup/build",
+        "2-4 observation",
+        "verdict/evidence",
+    ),
+}
+
 INTERNAL_READONLY_CLOSEOUT_PATTERNS = (
     (re.compile(r"\bself[- ]check\b", re.IGNORECASE), "Self-Check"),
     (re.compile(r"\bregistry (?:candidate )?scan\b", re.IGNORECASE), "registry scan"),
@@ -430,6 +458,20 @@ def validate_instruction_budgets(root: Path) -> tuple[list[str], list[str]]:
     return errors, warnings
 
 
+def validate_playtest_focus_contract(root: Path) -> list[str]:
+    errors: list[str] = []
+    for rel, required_phrases in PLAYTEST_FOCUS_CONTRACT.items():
+        path = root / rel
+        if not path.exists():
+            errors.append(f"{rel}: missing playtest focus contract surface")
+            continue
+        text = path.read_text(encoding="utf-8").lower()
+        missing = [phrase for phrase in required_phrases if phrase.lower() not in text]
+        if missing:
+            errors.append(f"{rel}: missing playtest focus contract phrase(s): {', '.join(missing)}")
+    return errors
+
+
 def validate_active_state_checkpoint_text(rel: Path, text: str, exempt: bool = False) -> list[str]:
     if exempt or ACTIVE_STATE_PATH not in text:
         return []
@@ -639,6 +681,7 @@ def main() -> int:
         budget_errors, budget_warnings = validate_instruction_budgets(root)
         errors.extend(budget_errors)
         warnings.extend(budget_warnings)
+        errors.extend(validate_playtest_focus_contract(root))
     if args.kind == "skills":
         errors.extend(validate_skills(root, args.require_present))
     if args.kind == "agents":

AGENTS.md

@@ -42,6 +42,9 @@ verification-first implementation.
   `Next action:` then `1. (Recommended) [action label] - [brief reason /
   command]`. Base that next action on the `## Session Worklist` when
   `production/session-state/active.md` exists. The user can reply with `1`.
+- When the next action or owed verification is a user-owned playtest, include
+  `Playtest focus:` with the hypothesis, setup/build if known, 2-4 observation
+  prompts, and the verdict/evidence the user should return.
 
 ## Available Codex Role Agents
 

CHANGELOG.md

@@ -4,6 +4,19 @@
 
 - Nothing yet.
 
+## v0.4.4 - 2026-07-09
+
+- Added a user-owned playtest focus contract so closeouts and owed verification
+  must include a `Playtest focus:` brief instead of generic playtest requests.
+- Updated session continuity guidance to preserve the playtest hypothesis,
+  setup/build, observation prompts, and requested verdict/evidence in
+  `## Session Worklist` entries.
+- Updated `$playtest-report` templates and routing so new reports and follow-up
+  playtests carry a concrete hypothesis while leaving game-feel and balance
+  verdicts with the user.
+- Added runtime validation to keep the playtest-focus rule present in root
+  instructions, continuity docs, and the playtest-report workflow.
+
 ## v0.4.3 - 2026-07-08
 
 - Tightened the role-agent delegation contract so invoking a CCGS skill is the

README.md

@@ -4,7 +4,7 @@ Codex Game Studios turns a game repository into a Codex-native indie studio
 workflow: 49 role agents, 77 repo-local skills, verification-first handoffs, and
 Godot-first production guidance for small teams building playable slices.
 
-Current package version: `0.4.2`.
+Current package version: `0.4.4`.
 
 This project is an unofficial Codex-native port of
 [Donchitos/Claude-Code-Game-Studios](https://github.com/Donchitos/Claude-Code-Game-Studios),
@@ -29,8 +29,16 @@ surfaces with Codex-native agents, skills, hooks, rules, and install behavior.
 
 ## Current Status
 
-The current release line is `v0.4.2`. It includes:
+The current release line is `v0.4.4`. It includes:
 
+- User-owned playtest focus routing: when owed verification or the next action
+  is a manual playtest, closeouts include a `Playtest focus:` brief with the
+  hypothesis, setup/build, observation prompts, and verdict/evidence to return.
+- `$playtest-report` templates and follow-up routing now require concrete
+  hypotheses before sending the user back to play, while preserving the user's
+  ownership of game-feel and balance verdicts.
+- Runtime validation keeps the playtest-focus contract present in root
+  instructions, session-continuity docs, and the playtest-report workflow.
 - Active session-state checkpoint updates: after the user approves the
   underlying workflow artifact or decision, skills may update only
   `production/session-state/active.md` without asking a second "May I write?"