codex-workflows

A system for running repeatable workflows in Codex, where a workflow is a defined sequence of tasks that may be shared across multiple agents. It supports large fan-out jobs, reviews, sweeps, and multi-agent coordination.

In plain words
What is it for?
Use it for release-difference reviews, security reviews, bug sweeps, adversarial checks, and other repeatable multi-agent jobs.
Why use it?
It removes the need to coordinate many similar coding-agent tasks manually. Saved workflows can be checked before they are run.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/robzilla1738/codex-workflows/codex-workflows
Any agent
npx skills add robzilla1738/Codex-Workflows --skill codex-workflows
Clone the repo
git clone --depth 1 https://github.com/robzilla1738/Codex-Workflows

Made for: Claude Code, Codex.

Per session 39 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 950 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00039 $0.00950
Opus 5 $0.00019 $0.00475
Sonnet 5 $0.00008 $0.00190
Haiku 4.5 $0.00004 $0.00095

Measured 2d ago against content hash dab80f7b2c83, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

codex-workflows scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Asks the agent to reveal its instructionslowSystem prompt leakage

Directions to print, repeat or translate the system prompt extract configuration the operator did not intend to expose.

- `promptSuffix` appends user-specific output instructions to every worker.

Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.

plugins/codex-workflows/skills/codex-workflows/SKILL.md · 78 lines

How it starts

The opening of the file, as written. The whole thing — 78 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Codex Workflows

Use this skill when the user asks to run a workflow, fan out many agents, perform release diff review, run a security or bug sweep, or save a repeatable multi-agent process.

Workflow

  1. Prefer existing saved workflows under .codex-workflows/workflows/ or the built-in workflows under workflows/.
  2. Validate or preview unfamiliar workflows with workflow_validate or workflow_preview before launch.
  3. Start the workflow through the codex-workflows MCP server when available.
  4. Expect workflow_run to open the live terminal dashboard automatically in the user's default terminal. Pass openTui: false only when the user asks for a headless run.
  5. Use adapter: "auto" for real Codex runs unless the user explicitly asks for sdk, exec, or simulate. Auto tries SDK first and falls back to codex exec when SDK cannot launch.
  6. When the user names subagent models, pass them as Codex model slugs:
    • model sets the default worker model and overrides workflow-file models.
    • reasoning sets default Codex reasoning effort.
    • modelMap maps phase id, agent id, or phase:agent to a model.
    • promptSuffix appends user-specific output instructions to every worker.
    • If the user gives shorthand such as 5.4-mini, use the full Codex slug if known, such as gpt-5.4-mini. The runtime validates model names before fanout and will return a clear error if the model is unavailable.
  7. For read-only review, do not pass approval: "deny". That denies workflow launch. Rely on read-only worker sandboxing and explicit no-mutation instructions in promptSuffix.
  8. Use the system-default/wide terminal defaults unless the user asks otherwise: terminalApp: "default", terminalColumns: 190, terminalRows: 42. The plugin accepts explicit terminal sizes up to 500 columns by 120 rows.
  9. Store normal bug-hunt run artifacts in Codex home by default. Use storageScope: "project" only when the user asks for project-local run logs or when saving a reusable workflow.
  10. If MCP is unavailable in an installed plugin, ask the user to restart Codex or reinstall/enable the plugin before running workflows. Use the repo-local pnpm cwf CLI only when working inside the codex-workflows source repo.
  11. Keep the main Codex thread concise. Poll status and summarize progress instead of pasting raw worker logs. The dashboard shows recent worker activity live, but exact token totals may remain pending until Codex emits usage metadata.
  12. For bug finding, default to the bounded workflows/bug-sweep.workflow.js. Use workflows/bug-sweep-deep.workflow.js only when the user explicitly asks for a larger or deeper fanout.
  13. For release review, default to workflows/release-diff-review.workflow.js.
  14. For security review, default to workflows/security-auth-review.workflow.js.
  15. Before write-capable workflows, state the sandbox mode and ask for approval.

Read the full file on GitHub · 78 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 78 lines · 39 tokens per session scan A dab80f7b2c83

Subscribe to this mod's changes

codex-workflows is a skill published in the GitHub repository robzilla1738/Codex-Workflows (63 stars, last pushed 2mo ago), licensed MIT. It adds 39 tokens to every session and 950 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 1 finding (asks the agent to reveal its instructions). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

crewai-multi-agent

Multi-agent orchestration framework for autonomous AI collaboration. Use when building teams of specialized agents working together on complex tasks, when you need role-based agent collaboration with memory, or for production workflows requiring sequential/hierarchical execution. Built without LangChain dependencies…

davila7/claude-code-templates · 61 tokens

opencode-ensemble

Use when coordinating multiple coding agents, delegating independent software work, managing OpenCode Ensemble teams, choosing teammate roles or models, reviewing teammate output, or deciding whether parallel execution is appropriate.

hueyexe/opencode-ensemble · 42 tokens

crewai-multi-agent

Multi-agent orchestration framework for autonomous AI collaboration. Use when building teams of specialized agents working together on complex tasks, when you need role-based agent collaboration with memory, or for production workflows requiring sequential/hierarchical execution. Built without LangChain dependencies…

OpenLAIR/dr-claw · 61 tokens

lark-bot-permissions

Grant/open Feishu permission scopes for a TaroCub Lark bot APP by driving the Feishu developer console in a browser — QR-login (send the QR to the user over the bridge), bulk-import the scope JSON, 申请开通, verify, and restart the instance. Use when the user wants to "开/加/申请 权限/scope" for a bot (e.g. 群里非@也回应 / Sheets /…

cloveric/tarocub · 157 tokens

scrapling

Web scraping that gets past anti-bot — HTTP fetch / stealth (Cloudflare bypass) / JS-render / spider crawl, via the scrapling CLI. Use when webextract is blocked or returns empty, or to scrape a protected/dynamic site or crawl many pages. Triggers: scrape / crawl a site, "scrape this page", "the site blocks bots /…

cloveric/tarocub · 161 tokens

tarocub-service-ops

Operate the TaroCub Lark bot fleet — restart, status, stop, start, logs, doctor — via its service manager. Use when asked to restart the fleet/bots, check whether the service is running or healthy, stop or start the bots, view bot logs, or diagnose the Lark service. Triggers: 重启车队 / 重启bot / restart the fleet, 服务状态 /…

cloveric/tarocub · 110 tokens