Agent skill

Codex Coding Plugin

by styler-ai in styler-ai/ProjectAtlas

Build, review, or fix ProjectAtlas plugin/runtime installer integration for Codex, Claude Code, and OpenCode, especially version convergence, stale ProjectAtlas cache repair, MCP config generation…

MITAuto-check passedTesting & QA

Install Codex Coding Plugin

skills CLI
$ npx skills add styler-ai/ProjectAtlas --skill codex-coding-plugin -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install styler-ai/ProjectAtlas codex-coding-plugin --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/styler-ai/ProjectAtlas.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/codex-coding-plugin .claude/skills/codex-coding-plugin && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
codex-coding-plugin
GitHub stars
438
Token cost
~1.8k tokens
SKILL.md length
837 words
Files
2
Skills in repo
2
Repo updated
First seen
Licence
MIT

At a glance

Build, review, or fix ProjectAtlas plugin/runtime installer integration for Codex, Claude Code, and OpenCode, especially version convergence, stale ProjectAtlas cache repair, MCP config generation…

  • Works in 9 steps: Treat the ProjectAtlas release tag and… → Never trust one version signal alone.… → Prefer official host commands for host… → …
  • Tasks that involve QA and bug reports
  • SKILL.md covers Goal, Workflow, Version Checks and Tests, plus 1 more section
  • Calls opencode, claude and codex

What it does

Codex Coding Plugin is an agent skill from styler-ai/ProjectAtlas. Build, review, or fix ProjectAtlas plugin/runtime installer integration for Codex, Claude Code, and OpenCode, especially version convergence, stale ProjectAtlas cache repair, MCP config generation, skill artifacts, host smoke tests, and fake-host tests for ProjectAtlas releases.

Its SKILL.md is about 1.8k tokens, which your agent loads only when the skill is triggered. The skill folder holds 2 other files (for example `agents/openai.yaml`).

It sits in Testing & QA, covering QA and bug reports. It works with Model Context Protocol, SQLite and Rust. The repository describes itself as: Every file not opened. Every folder not explored. Tokens saved. ProjectAtlas guides coding agents with purpose metadata and an intelligent code graph, reducing token costs by… The licence is MIT.

When your agent uses it

  • Tasks that involve QA and bug reports

Example prompts

  • “/codex-coding-plugin”

Workflow steps

9 steps, taken from the first numbered list in SKILL.md.

  1. Treat the ProjectAtlas release tag and plugin manifest versions as the release contract, but verify them against the native runtime with…
  2. Never trust one version signal alone. Check every ProjectAtlas surface that can keep stale data
  3. Prefer official host commands for host state. Inspect cache/source files only after the host reports the source path and after confirming…
  4. Make official ProjectAtlas paths work. If an official Codex marketplace/plugin cache is stale, repair it automatically with the supported…
  5. Preserve user-managed environments. If the marketplace/source is not clearly the official styler-ai/ProjectAtlas source, do not mutate it…
  6. Claude Code and OpenCode do not use the Codex marketplace/cache model in this repo. For them, verify the generated project-local config…
  7. Pin generated MCP configs to absolute verified runtime, DB, and config paths. Avoid PATH-only fallbacks for ProjectAtlas-installed configs.
  8. Compare canonical runtime paths, including symlink resolution where available. A shim path and resolved executable path must not create a…
  9. Keep installer output concrete: say which host, version, path, config, cache, registry entry, or workflow pin was verified, repaired…

What it can do on your machine

Read from SKILL.md and the folder at commit 5d526c2. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • opencode
    • claude
    • codex

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Codex Coding Plugin loads about 1.8k tokens when it runs. Until then it costs about 75 tokens; SKILL.md has 837 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~75
When it runs · the whole SKILL.md, loaded when a task matches
~1.8k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from styler-ai/ProjectAtlas at commit 5d526c2, republished under its MIT licence (© styler-ai). 837 words, ~1,836 tokens.

Download SKILL.mdSave it as .claude/skills/codex-coding-plugin/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.
name
codex-coding-plugin
description
Build, review, or fix ProjectAtlas plugin/runtime installer integration for Codex, Claude Code, and OpenCode, especially version convergence, stale ProjectAtlas cache repair, MCP config generation, skill artifacts, host smoke tests, and fake-host tests for ProjectAtlas releases.

ProjectAtlas Plugin Host Integration

Goal

Make ProjectAtlas releases converge to one verified version across the repository release tag, ProjectAtlas plugin manifests, native runtime binary, generated MCP configs, Codex marketplace/plugin cache, Codex global MCP registry, packaged ProjectAtlas skill files, Claude Code config, OpenCode config, and any old ProjectAtlas information that can survive in a host cache.

This is a repo-local ProjectAtlas skill. Do not generalize these rules into product claims for unrelated plugins. The point is to keep this repository's supported host integrations from reintroducing stale ProjectAtlas versions or paths.

Workflow

  1. Treat the ProjectAtlas release tag and plugin manifest versions as the release contract, but verify them against the native runtime with projectatlas --format json runtime-info before writing configs.
  2. Never trust one version signal alone. Check every ProjectAtlas surface that can keep stale data:
    • .codex-plugin/plugin.json, .claude-plugin/plugin.json, and opencode/opencode.json;
    • Codex plugin list version and, when exposed, the reported installed source path;
    • manifest and ProjectAtlas skill artifact under the reported Codex source path;
    • runtime runtime-info project, major version, capabilities, executable, and version;
    • generated MCP --require-version, DB path, config path, and final mcp command;
    • Codex global MCP registry runtime, DB/config, and version guard;
    • Claude Code .mcp.json config consumption;
    • OpenCode opencode.json config consumption;
    • downstream .github/workflows release pins that still point at old ProjectAtlas tags.
  3. Prefer official host commands for host state. Inspect cache/source files only after the host reports the source path and after confirming it is the official ProjectAtlas source.
  4. Make official ProjectAtlas paths work. If an official Codex marketplace/plugin cache is stale, repair it automatically with the supported remove/add, upgrade, or marketplace replacement flow, then re-read host state. Do not stop at a stale-cache warning unless the environment is intentionally managed or repair still cannot converge.
  5. Preserve user-managed environments. If the marketplace/source is not clearly the official styler-ai/ProjectAtlas source, do not mutate it; report a concrete user-managed skip.
  6. Claude Code and OpenCode do not use the Codex marketplace/cache model in this repo. For them, verify the generated project-local config with structured JSON parsing, then smoke the real host CLI when installed.
  7. Pin generated MCP configs to absolute verified runtime, DB, and config paths. Avoid PATH-only fallbacks for ProjectAtlas-installed configs.
  8. Compare canonical runtime paths, including symlink resolution where available. A shim path and resolved executable path must not create a false mismatch if they refer to the same verified runtime.
  9. Keep installer output concrete: say which host, version, path, config, cache, registry entry, or workflow pin was verified, repaired, skipped, or still mismatched.

Version Checks

Use a single small helper for each host surface so the installer checks data the same way in implementation and tests. The common ProjectAtlas bug is checking one fresh surface, then installing from another stale surface: for example, codex plugin list reports the expected version while the reported source path still contains an old manifest or old ProjectAtlas skill.

Required guard for Codex ProjectAtlas plugin repair:

text
if plugin_list.version == expected
and marketplace_ref == expected_tag
and reported_source_manifest.version == expected:
    no-op
else if official ProjectAtlas marketplace:
    remove/add or upgrade the plugin
    re-read plugin list and source manifest
    require both to match expected
else:
    skip mutation with clear user-managed message

Required guard for ProjectAtlas generated host configs:

text
runtime_info.version == expected
and command_path == canonical_verified_runtime
and args include "--require-version", expected, "--db", selected_db, "mcp"
and optional config path == selected_config
and host-specific fields match the supported shape

OpenCode host-specific fields are type = "local", enabled = true, command array shape, and project cwd. Claude Code uses an .mcp.json-compatible mcpServers.projectatlas shape and must not rely on cwd for project binding because DB/config arguments are absolute.

Show full SKILL.md (309 more words)Show less

Tests

Use fake host commands for deterministic installer tests and real host CLI smoke tests when installed locally. The minimum regression matrix is:

  • official marketplace, stale plugin list version -> repairs;
  • official marketplace, current plugin list version but stale reported source manifest -> repairs;
  • official marketplace, current plugin/source manifest but stale ProjectAtlas skill artifact -> repairs or fails before claiming convergence;
  • official marketplace, current plugin and source manifest -> no-op;
  • non-official marketplace -> no mutation;
  • failed repair -> restores when possible and reports the mismatched field;
  • generated MCP config launches the verified runtime;
  • generated config uses --require-version, selected DB/config, and final mcp;
  • generated config parsing uses structured JSON (ConvertFrom-Json, jq, or python3), not sed/regex extraction;
  • POSIX path comparison resolves existing runtime symlinks before declaring a mismatch;
  • OpenCode config requires type = "local", enabled = true, command array shape, and cwd;
  • Claude Code config requires the .mcp.json server shape and absolute DB/config binding.

For installer tests, assert exact command fragments for remove/add/update calls and inspect the generated config files, not just process success. For real local smoke, prefer:

bash
opencode --version
opencode debug config
claude --version
claude mcp get projectatlas

Summarize only the ProjectAtlas config fields from host debug output. Host debug commands may print unrelated local environment values; do not paste full raw output into issues, PRs, or release notes.

Release Gates

Before calling a ProjectAtlas host-plugin release stable, verify:

  • ProjectAtlas plugin manifest versions match the release tag;
  • runtime runtime-info reports the release version, ProjectAtlas identity, MCP capability, and TOON output;
  • installer writes project-local Codex-compatible, Claude Code, and OpenCode MCP configs with --require-version;
  • installer repairs official stale Codex marketplace/plugin cache, Codex ProjectAtlas skill artifact drift, and Codex global MCP registry drift when codex is available and not intentionally skipped;
  • installer verifies Claude Code and OpenCode generated configs against the verified runtime and host-specific schema;
  • local smoke covers OpenCode and Claude Code if installed on the machine;
  • release tests run on Windows and POSIX paths when both installer scripts exist;

© styler-ai, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 1 other file in skills/codex-coding-plugin of styler-ai/ProjectAtlas.

  • SKILL.md
  • agents/openai.yaml

Open the folder on GitHubat commit 5d526c2

Compare with similar skills

Codex Coding Plugin next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Codex Coding Plugin compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Codex Coding Plugin this skillstyler-ai/ProjectAtlas438—~1.8kAutomated safety check: PassMIT
Foremergenaw103/foremerge536—~2.4kAutomated safety check: PassApache-2.0
Memorywhalewuisabel-gif/MemWhale151—~765Automated safety check: PassMIT
Memorywhale Evidencewuisabel-gif/MemWhale151—~459Automated safety check: PassMIT
Memorywhale Debuggingwuisabel-gif/MemWhale151—~370Automated safety check: PassMIT
OpenWork Desktop CDP Driverdifferent-ai/openwork24k—~465Automated safety check: PassCustom licence

Similar skills

  • Foremerge

    naw103/foremerge

    Coordinate parallel coding agents with Foremerge's local Git-compatible CLI and MCP server.

    536 GitHub stars~2.4k tokensUpdated 3 days ago
    Agent WorkflowsAuto-check passed
  • Memorywhale

    wuisabel-gif/MemWhale

    Query and write durable debugging memory recorded by MemoryWhale.

    151 GitHub stars~765 tokensUpdated yesterday
    Agent WorkflowsAuto-check passed
  • Memorywhale Evidence

    wuisabel-gif/MemWhale

    Use MemoryWhale debugging evidence when the user requests recall or a recurring failure may have relevant recorded history.

    151 GitHub stars~459 tokensUpdated yesterday
    Agent WorkflowsAuto-check passed
  • Memorywhale Debugging

    wuisabel-gif/MemWhale

    MemoryWhale debugging; compiler failures; terminal diagnostics.

    151 GitHub stars~370 tokensUpdated yesterday
    DevelopmentAuto-check passed
  • OpenWork Desktop CDP Driver

    different-ai/openwork

    Drives a running OpenWork desktop window over CDP from the shell to evaluate JS, take screenshots, start sessions and send prompts for hand checks.

    24k GitHub stars~465 tokensUpdated today
    Testing & QAAuto-check passed
  • OpenCode QA Toolkit

    code-yeongyu/oh-my-openagent

    Tests the opencode coding agent itself: its CLI, server, plugin hooks and events, the terminal UI under tmux, and its SQLite session database, using tested helper scripts.

    70k GitHub stars~2.9k tokensUpdated today
    Testing & QAAuto-check passed

More from styler-ai/ProjectAtlas

  • Projectatlas

    styler-ai/ProjectAtlas

    Use ProjectAtlas before broad source reads, preferring the installed short atlas CLI for an exact checkout and MCP for registered worktree routing, compact session briefs, or federated graph evidence.

    438 GitHub stars~9.2k tokensUpdated yesterday
    Auto-check passed

Categories

Questions about Codex Coding Plugin

What does Codex Coding Plugin do?

Build, review, or fix ProjectAtlas plugin/runtime installer integration for Codex, Claude Code, and OpenCode, especially version convergence, stale ProjectAtlas cache repair, MCP config generation…. Codex Coding Plugin is an agent skill from styler-ai/ProjectAtlas. Build, review, or fix ProjectAtlas plugin/runtime installer integration for Codex, Claude Code, and OpenCode, especially version convergence, stale ProjectAtlas cache repair, MCP config generation, skill artifacts, host smoke tests, and fake-host tests for ProjectAtlas releases.

When should I use Codex Coding Plugin?

Codex Coding Plugin fits situations like: tasks that involve QA and bug reports.

How do I install Codex Coding Plugin in Claude Code?

Run `npx skills add styler-ai/ProjectAtlas --skill codex-coding-plugin -a claude-code`. Or copy the skill folder (skills/codex-coding-plugin in styler-ai/ProjectAtlas) into .claude/skills/codex-coding-plugin in your project. Claude Code loads it when a task matches its description.

How do I install Codex Coding Plugin in Codex?

Run `npx skills add styler-ai/ProjectAtlas --skill codex-coding-plugin -a codex`. Or copy the skill folder (skills/codex-coding-plugin in styler-ai/ProjectAtlas) into .agents/skills/codex-coding-plugin in your project. Codex loads it when a task matches its description.

Can I use Codex Coding Plugin in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add styler-ai/ProjectAtlas --skill codex-coding-plugin -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/codex-coding-plugin, .gemini/skills/codex-coding-plugin, .github/skills/codex-coding-plugin and .opencode/skills/codex-coding-plugin in your project.

What does Codex Coding Plugin need to run?

Going by SKILL.md and its folder, Codex Coding Plugin needs the command-line tools its instructions call (opencode, claude and codex).

Does Codex Coding Plugin access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Codex Coding Plugin safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Codex Coding Plugin use?

Codex Coding Plugin is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Codex Coding Plugin use?

About 1.8k tokens (SKILL.md is roughly 7.3k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Codex Coding Plugin?

Skills that share tags, products or a category with Codex Coding Plugin: Foremerge (naw103/foremerge, 536 stars), Memorywhale (wuisabel-gif/MemWhale, 151 stars), Memorywhale Evidence (wuisabel-gif/MemWhale, 151 stars) and Memorywhale Debugging (wuisabel-gif/MemWhale, 151 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Codex Coding Plugin?

styler-ai (a GitHub user) maintains it in styler-ai/ProjectAtlas, which has 438 GitHub stars. The repository holds 2 skills in this directory. The repository was last updated on October 6, 2026.

Source: styler-ai/ProjectAtlas on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.