Fix self-repair status command contract

apps/dashboard/smoke/action-packet-smoke.ts

@@ -87,7 +87,7 @@ const legacyFallbackPacket = buildActionPacket({
   gateFallbackDecision: "保持 status inspection；补 project_asset 后再恢复 delivery。",
   boundary: "This is a legacy/raw fallback; do not infer owner, gate, or stop condition authority.",
   safePathLabel: "Legacy status inspection",
-  command: "loopx status --goal-id legacy-status-only",
+  command: "loopx diagnose --goal-id legacy-status-only --limit 20",
   projectNextAction: "Continue from raw status field.",
   projectStopCondition: "Stop before any delivery claim.",
   projectAssetSource: "legacy_raw_fallback",
@@ -111,7 +111,7 @@ const focusWaitPacket = buildActionPacket({
   gateFallbackDecision: "继续保持 focus wait；有新 owner evidence、clean baseline 或外部 eval 后再恢复 delivery。",
   boundary: "这不是 delivery approval；项目 Agent 只做 status/history inspection，不执行交付路径、写入、reward append 或生产动作。",
   safePathLabel: "Status/history inspection only",
-  command: "loopx --registry ./examples/registry.example.json --runtime-root ./tmp/runtime status --goal-id focus-wait-owner-blocker",
+  command: "loopx --registry ./examples/registry.example.json --runtime-root ./tmp/runtime diagnose --goal-id focus-wait-owner-blocker --limit 20",
 });
 
 assert(focusWaitPacket.includes("目标：focus-wait-owner-blocker"), "missing focus-wait goal id");
@@ -135,7 +135,7 @@ const platformMigrationNoEvidencePacket = buildActionPacket({
   boundary: "Use only sanitized status/history/material counts; do not read private evidence, internal links, raw paths, or production state.",
   durableRecordRule: null,
   safePathLabel: "No-evidence status/packet sanity",
-  command: "loopx status --goal-id platform-migration-material-registry --limit 20",
+  command: "loopx diagnose --goal-id platform-migration-material-registry --limit 20",
   quotaShortLine: "Eligible; 0/1440 slots",
   authorityShortLine: "entries 0/3; topics 3; materials 6; repos 2; owner review 1; stale 1; risk medium",
   projectOwner: "codex",

examples/install-local-smoke.py

@@ -178,11 +178,16 @@ def main() -> int:
         self_repair_text = " ".join(self_repair_skill.read_text(encoding="utf-8").split())
         for phrase in (
             "Build a compact evidence packet",
+            "loopx --format json diagnose --goal-id <goal-id>",
+            "loopx --format json status --limit 20",
+            "intentionally a registry/dashboard view rather than a goal-filtered command",
+            "registry-declared active state file",
             "references/repair-patterns.md",
             "Repair at the lowest durable layer",
             "Do not solve contradictory payloads by guessing",
         ):
             assert phrase in self_repair_text, phrase
+        assert "status --goal-id <goal-id>" not in self_repair_text, self_repair_text
         self_repair_patterns = (
             codex_home
             / "skills"
@@ -192,6 +197,7 @@ def main() -> int:
         )
         self_repair_patterns_text = self_repair_patterns.read_text(encoding="utf-8")
         assert "`boundary_projection_gap`" in self_repair_patterns_text, self_repair_patterns_text
+        assert "`skill_cli_contract_drift`" in self_repair_patterns_text, self_repair_patterns_text
         assert "`tiny_turn_under_delivery`" in self_repair_patterns_text, self_repair_patterns_text
         assert (
             codex_home / "skills" / "loopx-self-repair" / "agents" / "openai.yaml"

examples/review-packet-cli-smoke.py

@@ -735,7 +735,7 @@ def assert_missing_project_asset_review_packet_fallback() -> None:
                     "waiting_on": "codex",
                     "severity": "action",
                     "recommended_action": "Continue only through raw status fallback.",
-                    "agent_command": "loopx status --goal-id legacy-status-only",
+                    "agent_command": "loopx diagnose --goal-id legacy-status-only --limit 20",
                     "source": "latest_run",
                 }
             ]

examples/review-packet-smoke.py

@@ -126,7 +126,7 @@ def build_sanitized_focus_wait_packet() -> str:
     goal_id = "focus-wait-owner-blocker"
     status_command = (
         "loopx --registry ./examples/registry.example.json "
-        "--runtime-root ./tmp/runtime status --goal-id focus-wait-owner-blocker"
+        "--runtime-root ./tmp/runtime diagnose --goal-id focus-wait-owner-blocker --limit 20"
     )
     return "\n".join(
         [

loopx/doctor.py

@@ -39,6 +39,9 @@
     ),
     "loopx-self-repair": (
         "Build a compact evidence packet",
+        "loopx --format json diagnose --goal-id <goal-id>",
+        "loopx --format json status --limit 20",
+        "registry-declared active state file",
         "references/repair-patterns.md",
         "Repair at the lowest durable layer",
     ),

skills/loopx-self-repair/SKILL.md

@@ -16,14 +16,17 @@ not only an apology or a one-off explanation.
 
    ```bash
    git status --short --branch
-   loopx --format json status --goal-id <goal-id>
-   loopx --format json quota should-run --goal-id <goal-id>
+   loopx --format json diagnose --goal-id <goal-id>
+   loopx --format json status --limit 20
+   loopx --format json quota should-run --goal-id <goal-id> [--agent-id <agent-id>]
    loopx --format json history --goal-id <goal-id> --limit 5
    ```
 
-   Also inspect the project-local registry and
-   `.codex/goals/<goal-id>/ACTIVE_GOAL_STATE.md` when relevant. Use the shared
-   global registry for heartbeat/quota truth.
+   `status` is intentionally a registry/dashboard view rather than a
+   goal-filtered command; use `diagnose --goal-id` for the goal-specific agent
+   evidence packet. Also inspect the project-local registry and the
+   registry-declared active state file when relevant. Use the shared global
+   registry for heartbeat/quota truth.
 3. **Classify the failure.** Read
    `references/repair-patterns.md` and match the symptoms to a known pattern.
    If no pattern fits, add one after the fix.

skills/loopx-self-repair/references/repair-patterns.md

@@ -22,6 +22,7 @@ teaches a reusable control-plane lesson.
 | `status_projection_history_neutral_gap` | `quota should-run` reports no user action or quiet monitor behavior, but `status` / `diagnose --limit N` misreports a stale controller/user gate or older connected-without-run state after recent monitor/spend/readiness events. | quota payload, diagnose/status payload, recent history classifications, neutral-run classification sets, UI `--limit`, latest meaningful state run. | Status/history/quota disagree on which runs are status-neutral, or a short display limit is reused as the control-plane reasoning window. | Share the neutral classification contract across history/status/quota and reason over a wider internal state window before trimming displayed runs; add a regression with consecutive neutral runs before a meaningful state transition. |
 | `monitor_replan_noop_loop` | Recent runs are monitor-only, replan, or self-repair-adjacent, but the same monitor/action recommendation keeps returning and no runnable todo, blocker, successor, supersede, user gate, capability change, or monitor-target change appears. | quota payload, interaction contract, todo summaries, recent history classifications, latest replan ACK, monitor target, active state todos. | Self-repair/replan was recorded as activity but did not change the machine-visible work frontier. | Require a repair delta contract: classify no-delta repairs as no-ops, then add a blocker, supersede stale monitor work, create a successor runnable todo, or record an explicit watch-lane continuation with expiry. |
 | `scheduler_liveness_backoff_gap` | `quota should-run` says `should_run=false`, `quiet_noop_allowed=true`, no spend, or `automation_action=keep_active_quiet`, but the host keeps polling at the bootstrap cadence for hours; Codex CLI TUI or Claude Code `/loop` also keeps repeating unchanged checks. | quota payload `automation_liveness`, `interaction_contract`, `heartbeat_recommendation`, host automation cadence, CLI/Claude loop logs or statusline, recent no-spend poll history. | LoopX separated execution permission from user notification, but did not expose a machine-readable next-wakeup/backoff/final-check/exit policy for host runtimes. | Add or repair `quota should-run.scheduler_hint`; host runtimes must apply Codex App cadence backoff and CLI/Claude unchanged-poll final quota/replan check before self-stop, without quota spend. Prompt/docs should treat fixed 3-minute cadence as bootstrap only. |
+| `skill_cli_contract_drift` | A skill, installed release snapshot, generated prompt, smoke fixture, or agent handbook suggests a CLI form that the current parser rejects, such as `loopx status --goal-id`. | `loopx <subcommand> --help`, failing command stderr, `rg` over `skills/`, installed skill copies, release snapshot, generated prompts, and relevant smoke fixtures. | Process documentation drifted from the CLI contract, or a fixture preserved old command text without labeling it as legacy/non-executable. | Update the durable skill/source docs and installed skill copy, add or update a focused smoke that rejects the stale command in executable instructions, and keep legacy fixture strings only when the test explicitly asserts they are untrusted historical payload. |
 | `loop_surface_activation_gap` | A thread reports LoopX setup success because `register-agent` or `quota should-run --agent-id` works, but no Codex App heartbeat automation, Codex CLI `/goal`, or other host loop has been installed for that goal. | `upgrade-plan` / host-loop activation status, `$CODEX_HOME/automations/*/automation.toml`, generated `heartbeat-prompt`, target thread id, recent setup final answer, registry coordination fields. | The control-plane identity layer was treated as the host scheduler layer; setup success criteria stopped at registry/quota instead of proving `host_loop_activation.activated=true` or surfacing a concrete host-tool gate. | Project setup and agent registration must surface `host_loop_activation`; create/update the host loop from generated scoped `heartbeat-prompt`, or report the exact missing host capability. Do not claim "connected" as autonomous until the host loop is active. |
 | `tiny_turn_under_delivery` | Many heartbeats make only one small doc/prose edit despite quota permitting a coherent batch. | recent history durations, active priority stack, user feedback, todo completion rate. | Agent optimized for avoiding mistakes rather than delivering a bounded batch. | Run steering audit, choose a larger verifiable batch, and record why it remains within boundary. Consider cadence/budget policy updates if repeated. |
 | `todo_succession_gap` | A non-trivial feature todo is marked done after PR/smoke success, but rollout, product-path audit, docs, telemetry, benchmark proof, or operator-decision follow-up is only mentioned in prose or not recorded at all. | completed todo note/evidence, recent PR/run history, active `Agent Todo`, active `Next Action`, related design docs. | Agent treated an implementation slice as feature completion, or LoopX lacked a reminder to create the successor todo. | Keep the lifecycle simple: do not add many feature states. Complete the old todo and create the next concrete agent/user todo with `--next-agent-todo`/`todo add`, or record a compact no-follow-up rationale. |