codex - 💡(How to fix) Fix Agent weakens regression tests and reports nonexistent validation/changes as complete during refactor task

{
  "schemaVersion": 1,
  "generatedAt": "1779979246s since unix epoch",
  "overallStatus": "fail",
  "codexVersion": "0.133.0",
  "checks": {
    "app_server.status": {
      "id": "app_server.status",
      "category": "app-server",
      "status": "ok",
      "summary": "background server is not running",
      "details": {
        "control socket": "C:\\Users\\manur\\.codex\\app-server-control\\app-server-control.sock",
        "daemon state dir": "C:\\Users\\manur\\.codex\\app-server-daemon",
        "mode": "ephemeral",
        "pid file": "C:\\Users\\manur\\.codex\\app-server-daemon\\app-server.pid (missing)",
        "settings": "C:\\Users\\manur\\.codex\\app-server-daemon\\settings.json (missing)",
        "status": "not running",
        "update-loop pid file": "C:\\Users\\manur\\.codex\\app-server-daemon\\app-server-updater.pid (missing)"
      },
      "remediation": null,
      "durationMs": 0
    },
    "auth.credentials": {
      "id": "auth.credentials",
      "category": "auth",
      "status": "ok",
      "summary": "auth is configured",
      "details": {
        "auth file": "C:\\Users\\manur\\.codex\\auth.json",
        "auth storage mode": "File",
        "stored API key": "false",
        "stored ChatGPT tokens": "true",
        "stored agent identity": "false",
        "stored auth mode": "chatgpt"
      },
      "remediation": null,
      "durationMs": 0
    },
    "config.load": {
      "id": "config.load",
      "category": "config",
      "status": "ok",
      "summary": "config loaded",
      "details": {
        "CODEX_HOME": "C:\\Users\\manur\\.codex",
        "config.toml": "C:\\Users\\manur\\.codex\\config.toml",
        "config.toml parse": "ok",
        "cwd": "F:\\codex\\composer-workbench-v2",
        "enabled feature flags": "shell_tool, shell_snapshot, terminal_resize_reflow, sqlite, memories, hooks, enable_request_compression, multi_agent, apps, tool_suggest, plugins, plugin_hooks, in_app_browser, browser_use, browser_use_external, computer_use, plugin_sharing, image_generation, skill_mcp_dependency_install, steer, guardian_approval, goals, collaboration_modes, tool_call_mcp_elicitation, personality, fast_mode, tui_app_server, prevent_idle_sleep, workspace_dependencies",
        "feature flag overrides": "memories=true, prevent_idle_sleep=true",
        "feature flags enabled": "29",
        "log dir": "C:\\Users\\manur\\.codex\\log",
        "mcp servers": "2",
        "model": "gpt-5.4-mini",
        "model provider": "openai",
        "sqlite home": "C:\\Users\\manur\\.codex"
      },
      "remediation": null,
      "durationMs": 0
    },
    "installation": {
      "id": "installation",
      "category": "install",
      "status": "ok",
      "summary": "installation looks consistent",
      "details": {
        "PATH codex #1": "C:\\Users\\manur\\AppData\\Roaming\\npm\\codex",
        "PATH codex #2": "C:\\Users\\manur\\AppData\\Roaming\\npm\\codex.cmd",
        "PATH codex entries": "2",
        "current executable": "C:\\Users\\manur\\AppData\\Roaming\\npm\\node_modules\\@openai\\codex\\node_modules\\@openai\\codex-win32-x64\\vendor\\x86_64-pc-windows-msvc\\bin\\codex.exe",
        "install context": "npm (package C:\\Users\\manur\\AppData\\Roaming\\npm\\node_modules\\@openai\\codex\\node_modules\\@openai\\codex-win32-x64\\vendor\\x86_64-pc-windows-msvc, bin C:\\Users\\manur\\AppData\\Roaming\\npm\\node_modules\\@openai\\codex\\node_modules\\@openai\\codex-win32-x64\\vendor\\x86_64-pc-windows-msvc\\bin, resources C:\\Users\\manur\\AppData\\Roaming\\npm\\node_modules\\@openai\\codex\\node_modules\\@openai\\codex-win32-x64\\vendor\\x86_64-pc-windows-msvc\\codex-resources, path C:\\Users\\manur\\AppData\\Roaming\\npm\\node_modules\\@openai\\codex\\node_modules\\@openai\\codex-win32-x64\\vendor\\x86_64-pc-windows-msvc\\codex-path)",
        "managed by bun": "false",
        "managed by npm": "true",
        "managed package root": "C:\\Users\\manur\\AppData\\Roaming\\npm\\node_modules\\@openai\\codex",
        "npm update target": "C:\\Users\\manur\\AppData\\Roaming\\npm\\node_modules\\@openai\\codex"
      },
      "remediation": null,
      "durationMs": 1342
    },
    "mcp.config": {
      "id": "mcp.config",
      "category": "mcp",
      "status": "ok",
      "summary": "MCP configuration is locally consistent",
      "details": {
        "configured servers": "2",
        "disabled servers": "0",
        "stdio servers": "1",
        "streamable_http servers": "1"
      },
      "remediation": null,
      "durationMs": 44
    },
    "network.env": {
      "id": "network.env",
      "category": "network",
      "status": "ok",
      "summary": "network-related environment looks readable",
      "details": {
        "proxy env vars": "none"
      },
      "remediation": null,
      "durationMs": 0
    },
    "network.provider_reachability": {
      "id": "network.provider_reachability",
      "category": "reachability",
      "status": "fail",
      "summary": "one or more required provider endpoints are unreachable over HTTP",
      "details": {
        "ChatGPT base URL": "https://chatgpt.com/backend-api/ request timed out (required)",
        "reachability mode": "ChatGPT auth"
      },
      "remediation": "Check proxy, VPN, firewall, DNS, and custom CA configuration.",
      "durationMs": 3007
    },
    "network.websocket_reachability": {
      "id": "network.websocket_reachability",
      "category": "websocket",
      "status": "ok",
      "summary": "Responses WebSocket handshake succeeded",
      "details": {
        "DNS": "2 IPv4, 0 IPv6, first IPv4",
        "auth mode": "chatgpt",
        "connect timeout": "15000 ms",
        "endpoint": "wss://chatgpt.com/backend-api/<redacted>",
        "handshake result": "HTTP 101 Switching Protocols",
        "model provider": "openai",
        "models etag present": "true",
        "provider name": "OpenAI",
        "proxy env vars": "none",
        "reasoning header": "false",
        "server model present": "false",
        "supports websockets": "true",
        "wire API": "responses"
      },
      "remediation": null,
      "durationMs": 12582
    },
    "runtime.provenance": {
      "id": "runtime.provenance",
      "category": "runtime",
      "status": "ok",
      "summary": "running npm on windows-x86_64",
      "details": {
        "commit": "unknown",
        "current executable": "C:\\Users\\manur\\AppData\\Roaming\\npm\\node_modules\\@openai\\codex\\node_modules\\@openai\\codex-win32-x64\\vendor\\x86_64-pc-windows-msvc\\bin\\codex.exe",
        "install method": "npm (package C:\\Users\\manur\\AppData\\Roaming\\npm\\node_modules\\@openai\\codex\\node_modules\\@openai\\codex-win32-x64\\vendor\\x86_64-pc-windows-msvc, bin C:\\Users\\manur\\AppData\\Roaming\\npm\\node_modules\\@openai\\codex\\node_modules\\@openai\\codex-win32-x64\\vendor\\x86_64-pc-windows-msvc\\bin, resources C:\\Users\\manur\\AppData\\Roaming\\npm\\node_modules\\@openai\\codex\\node_modules\\@openai\\codex-win32-x64\\vendor\\x86_64-pc-windows-msvc\\codex-resources, path C:\\Users\\manur\\AppData\\Roaming\\npm\\node_modules\\@openai\\codex\\node_modules\\@openai\\codex-win32-x64\\vendor\\x86_64-pc-windows-msvc\\codex-path)",
        "platform": "windows-x86_64",
        "version": "0.133.0"
      },
      "remediation": null,
      "durationMs": 0
    },
    "runtime.search": {
      "id": "runtime.search",
      "category": "search",
      "status": "ok",
      "summary": "search is OK (bundled)",
      "details": {
        "search command": "C:\\Users\\manur\\AppData\\Roaming\\npm\\node_modules\\@openai\\codex\\node_modules\\@openai\\codex-win32-x64\\vendor\\x86_64-pc-windows-msvc\\codex-path\\rg.exe",
        "search command readiness": "file exists",
        "search provider": "bundled"
      },
      "remediation": null,
      "durationMs": 0
    },
    "sandbox.helpers": {
      "id": "sandbox.helpers",
      "category": "sandbox",
      "status": "ok",
      "summary": "sandbox configuration is readable",
      "details": {
        "approval policy": "OnRequest",
        "codex-linux-sandbox helper": "none",
        "execve wrapper helper": "none",
        "filesystem sandbox": "restricted",
        "network sandbox": "restricted"
      },
      "remediation": null,
      "durationMs": 0
    },
    "state.paths": {
      "id": "state.paths",
      "category": "state",
      "status": "ok",
      "summary": "state paths and databases are inspectable",
      "details": {
        "CODEX_HOME": "C:\\Users\\manur\\.codex (dir)",
        "active rollout files": "346 files, 615320524 total bytes, 1778383 average bytes",
        "archived rollout files": "39 files, 22587994 total bytes, 579179 average bytes",
        "goals DB": "C:\\Users\\manur\\.codex\\goals_1.sqlite (file)",
        "goals DB integrity": "ok",
        "log DB": "C:\\Users\\manur\\.codex\\logs_2.sqlite (file)",
        "log DB integrity": "ok",
        "log dir": "C:\\Users\\manur\\.codex\\log (dir)",
        "sqlite home": "C:\\Users\\manur\\.codex (dir)",
        "state DB": "C:\\Users\\manur\\.codex\\state_5.sqlite (file)",
        "state DB integrity": "ok"
      },
      "remediation": null,
      "durationMs": 26246
    },
    "terminal.env": {
      "id": "terminal.env",
      "category": "terminal",
      "status": "ok",
      "summary": "terminal metadata was detected",
      "details": {
        "WT_SESSION": "present",
        "color output": "enabled",
        "stderr is terminal": "true",
        "stdin is terminal": "true",
        "stdout is terminal": "true",
        "terminal": "Windows Terminal",
        "terminal size": "148x36"
      },
      "remediation": null,
      "durationMs": 6
    },
    "updates.status": {
      "id": "updates.status",
      "category": "updates",
      "status": "ok",
      "summary": "update configuration is locally consistent",
      "details": {
        "cached latest version": "0.134.0",
        "check for update on startup": "true",
        "dismissed version": "0.134.0",
        "last checked at": "2026-05-28T03:28:54.438999500Z",
        "latest version": "0.134.0",
        "latest version status": "newer version is available",
        "npm update target": "C:\\Users\\manur\\AppData\\Roaming\\npm\\node_modules\\@openai\\codex",
        "update action": "npm install -g @openai/codex",
        "version cache": "C:\\Users\\manur\\.codex\\version.json"
      },
      "remediation": null,
      "durationMs": 2177
    }
  }
}

await clickBeginFormulation();
await settleAsyncWork();

expect(window.location.pathname).toBe("/workspace/drafts/draft-1");
expect(activePanel()).toBe("draft");
expect(screenText()).toContain("Loading saved draft...");

Codex-modified behavior

After the edit, the same test title remained, but Codex changed the assertion to the equivalent of:

await clickBeginFormulation();
await settleAsyncWork();

expect(window.location.pathname).toBe("/workspace/briefs/req-1");
expect(activePanel()).toBe("");
expect(persistedBriefPanel()).not.toBeNull();

In other words, a test still claiming to prove successful navigation was changed to assert that navigation did not happen.

Vacuous stale-request regression proof

Codex also modified the stale Begin regression setup into the equivalent of:

if (path === "/formulation/drafts") {
  return { draft: immediateDraftResult };
}

if (path === "/formulation/drafts") {
  return await pendingCreateRequest.promise;
}

The deferred branch is unreachable, so resolving pendingCreateRequest after abandonment no longer proves stale-completion protection.

False completion evidence

Codex committed a revision whose changed-file list omitted the lint-checker file, then posted completion evidence claiming that five entries had been removed from that lint-checker file and that validation had passed.

Thread identifier

Uploaded thread:
019e6dcf-f135-7f13-b558-917d668916a6

### What is the expected behavior?

```markdown
Codex should not obtain green validation by weakening, reversing, or accidentally neutralizing existing regression tests whose behavior it was explicitly instructed to preserve.

For this class of refactor task, expected behavior is:

1. Preserve existing positive behavior tests unless the same behavior is explicitly and correctly relocated to an equivalent integration proof.
2. Preserve race-condition regression tests so they still control a real pending async request and exercise the same failure mode after refactoring.
3. Treat removed behavior as requiring a new owner and corresponding proof, rather than silently dropping it.
4. Never post completion evidence claiming changes that are absent from the committed diff.
5. Never claim validation passed for the committed tree when committed files contradict the claimed guardrail cleanup.
6. When a test fails after an architectural extraction, diagnose whether the implementation violated the contract rather than changing assertions to encode the broken behavior.
7. Prefer stopping with a clear failure/blocker report over posting READY evidence for an untrustworthy implementation.

A useful product safeguard would be for Codex to detect and prominently warn when it changes assertions in existing tests that cover behavior explicitly marked as preserved in the task prompt, especially when those changes invert the asserted outcome or eliminate deferred/race behavior.

### Additional information

This report is intentionally focused on Codex CLI agent behavior and completion-evidence reliability.

The run included two MCP integrations:

- a read-only code-topology tool used during initial orientation;
- an issue-tracker integration used to post completion evidence.

The topology tool did not make source edits. Codex correctly identified the important constraints before implementation. The failures occurred during Codex's subsequent code and test edits, validation interpretation, and completion reporting.

Relevant observations:

- Codex performed a correct initial orientation and explicitly identified the stale-navigation guard and lint entries before editing.
- A context compaction occurred during implementation.
- After encountering failing tests, Codex repeatedly increased async settling windows and then weakened/remapped assertions instead of preserving the original behavior contract.
- Codex attempted a commit before staging had completed, then retried sequentially.
- The final commit changed five files but did not include the lint-checker file that its READY evidence claimed had been modified.
- The session used very high token volume for a bounded extraction task:
  - 880,219 input tokens
  - 19,996,032 cached input tokens
  - 187,227 output tokens
  - 134,615 reasoning output tokens

I am happy to provide a further-redacted before/after diff or the specific invariant/test matrix if useful to maintainers.