investigation-codex-vscode-extension
This skill should be used when investigating why the Codex VS Code extension fails to spawn or connect to a local Codex agent, especially after updates or CLI changes.
Best use case
investigation-codex-vscode-extension is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
This skill should be used when investigating why the Codex VS Code extension fails to spawn or connect to a local Codex agent, especially after updates or CLI changes.
Teams using investigation-codex-vscode-extension should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/investigation-codex-vscode-extension/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How investigation-codex-vscode-extension Compares
| Feature / Agent | investigation-codex-vscode-extension | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
This skill should be used when investigating why the Codex VS Code extension fails to spawn or connect to a local Codex agent, especially after updates or CLI changes.
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
Related Guides
SKILL.md Source
# Investigation: Codex VS Code Extension
*[Created by Codex: 019be564-39d9-7b12-a4cf-fb7a7f1af223 2026-01-22]*
## Purpose
Provide a fast, repeatable workflow to diagnose why the Codex VS Code extension cannot spawn or connect to the local `codex` app-server.
## When to Use
Use when a user reports that Codex in VS Code will not start, errors on spawn, or fails after an extension or CLI update.
## Warning Flags (Check First)
- **Non-default CLI override is set**: If `"chatgpt.cliExecutable"` is configured, the extension will run that binary instead of the bundled 45MB `codex`.
- **Assume “custom build” risk**: An override often points at a locally built or modified Codex repo; the user must know exactly what it is and why it differs.
- **Always surface it explicitly**: State the resolved `codex` path and `codex --version` in the RCA, and ask the user to confirm the repo/build they intended to run.
## Incident Example (2026-01-22)
- Symptom: “Cannot spawn a Codex agent in VS Code” after updating the extension to `openai.chatgpt` `0.4.62`.
- Root cause: The extension started `codex app-server --analytics-default-enabled`, but the user had `"chatgpt.cliExecutable"` pinned to a `codex-cli 0.77.0` binary that did not recognize that flag, so the process exited with code `2`.
- Evidence: `~/Library/Application Support/Code/logs/.../window*/exthost/openai.chatgpt/Codex.log` contained `error: unexpected argument '--analytics-default-enabled' found`.
- Fix: Remove or update `"chatgpt.cliExecutable"` to a compatible `codex` binary (or let the extension use its bundled one), then reload VS Code.
## Workflow (Non-Invasive)
1. Identify the *actual* `codex` binary VS Code is running (this is frequently the root cause):
- Check for `"chatgpt.cliExecutable"` in:
- `~/Library/Application Support/Code/User/settings.json`
- `<repo>/.vscode/settings.json`
- If set, treat it as a **high-priority hypothesis**: VS Code is running a non-default `codex` (often a custom build).
- If set, tell the user the resolved path and ask them to confirm which repo/build that binary came from.
2. Identify extension version and entrypoint:
- `ls ~/.vscode/extensions | rg -i 'openai.chatgpt'`
- `cat ~/.vscode/extensions/<ext>/package.json` (read `main`)
- `rg -n -- 'spawn|CLI_EXECUTABLE|codex|app-server' ~/.vscode/extensions/<ext>/out/extension.js`
3. Determine which `codex` binary the extension uses:
- If `"chatgpt.cliExecutable"` is set, it wins.
- Otherwise, the fallback is typically: `~/.vscode/extensions/<ext>/bin/<platform>/codex`.
4. Pull extension logs (most authoritative):
- `~/Library/Application Support/Code/logs/<latest>/window*/exthost/openai.chatgpt/Codex.log`
- `.../exthost/exthost.log` for activation and stack traces.
- Prioritize explicit CLI errors (unknown flags, `ENOENT`, `EACCES`, exit code).
5. Correlate failure with CLI arguments:
- Compare args in `extension.js` with `codex app-server --help` for the selected binary.
- If flags mismatch, conclude the extension is running an older CLI (often via `chatgpt.cliExecutable`).
6. Run sanity checks:
- Compare `codex --version` for (a) terminal PATH (b) configured `chatgpt.cliExecutable` target (c) extension-bundled binary.
- Verify executable permissions or quarantine status on macOS when `EACCES` appears.
7. Draft RCA and minimal fix:
- Cite the exact log line and file path.
- Provide the smallest fix: remove/update `chatgpt.cliExecutable`, point to newer binary, reload window.
## Common Root Causes
- `chatgpt.cliExecutable` pinned to an older CLI missing new flags.
- Extension update changed CLI arguments but local binary stayed old.
- Missing or non-executable bundled binary after update.
- macOS quarantine or code signing issues (`EACCES`).
## Output Expectations
- Provide a short RCA (1–3 bullets) and minimal fix steps.
- Avoid invasive code changes unless explicitly requested.Related Skills
vvvv-editor-extensions
Helps create vvvv gamma editor extensions — .HDE.vl file naming, Command node registration with keyboard shortcuts, SkiaWindow/SkiaWindowTopMost window types, docking with WindowFactory, and API access to hovered/selected nodes via VL.Lang Session nodes. Use when building editor plugins, custom tooling windows, or automating editor workflows.
reflect-codex-skills
Generate reflections for past Codex session histories using the Reflection CLI. Use when asked to summarize or reflect on previous Codex conversations, list projects/sessions, filter by date or session id, or refresh cached reflections from ~/.codex/sessions.
cursor-to-vscode-migration
Migrate editor setup from Cursor to VS Code. Cleans old VS Code data, copies settings/keybindings from Cursor, removes Cursor-specific config keys, and installs all extensions fresh. Use when switching from Cursor to VS Code.
copilot-vscode
VS Code Copilot platform reference: .agent.md all frontmatter fields, built-in tool names, agent types, context engineering, subagents, handoffs, MCP config, SKILL.md integration, custom instructions. Load when working with custom agents, tool lists, context strategy, or any VS Code Copilot platform feature.
codex-code-review
Perform comprehensive code reviews using OpenAI Codex CLI. This skill should be used when users request code reviews, want to analyze diffs/PRs, need security audits, performance analysis, or want automated code quality feedback. Supports reviewing staged changes, specific files, entire directories, or git diffs.
chrome-extension-init-skill
Initialize a Chrome/Chromium browser extension project from an empty folder by having the user choose a scaffolding toolchain (Plasmo, WXT, or Vite + CRXJS), then running the official interactive CLI (e.g., `npm create plasmo`, `npx wxt@latest init`, `npm create crxjs@latest`) to scaffold the project, and finishing with git init + first commit. Use when a user asks to start or bootstrap a browser extension from scratch.
browser-extension-builder
Expert in building browser extensions that solve real problems - Chrome, Firefox, and cross-browser extensions. Covers extension architecture, manifest v3, content scripts, popup UIs, monetization ...
axiom-extensions-widgets-ref
Use when implementing widgets, Live Activities, Control Center controls, or app extensions - comprehensive API reference for WidgetKit, ActivityKit, App Groups, and extension lifecycle for iOS 14+
openclaw-codex-oauth-proxy
OpenClaw 用 openai-codex(ChatGPT OAuth)跑 GPT-5.x Codex/Spark 的配置与排障(网关进程、代理、模型、thinking、JSON 输出)。
feature-investigation
[Investigation] Use when the user asks to investigate, explore, understand, explain, or analyze how an existing feature or logic works. Triggers on keywords like "how does", "explain", "what is the logic", "investigate", "understand", "where is", "trace", "walk through", "show me how".
codex-delegator-skill
Delegate complex tasks to Codex with structured prompts, tmux sessions, and logging utilities.
performing-malware-persistence-investigation
Systematically investigate all persistence mechanisms on Windows and Linux systems to identify how malware survives reboots and maintains access.