A file-based handoff process for giving a focused coding, research, or exploration task to an outside agent while Codex reviews the result.

Prepare a task packet, send it to Claude Code, Reasonix, or a manual worker, collect the output files, and review the result before calling the task complete.

It keeps delegated work bounded and reviewable, while reducing the risk of losing context or exposing secrets.

agent-task — Skill for Claude Code and Codex by leviuszen

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.

Any agent

npx skills add leviuszen/meaning-assurance --skill agent-task

Clone the repo

git clone --depth 1 https://github.com/leviuszen/meaning-assurance

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/skills/leviuszen/meaning-assurance/agent-task/github.svg)](https://agentmods.dev/skills/leviuszen/meaning-assurance/agent-task)

Your own site

<a href="https://agentmods.dev/skills/leviuszen/meaning-assurance/agent-task"><img src="https://agentmods.dev/badge/skills/leviuszen/meaning-assurance/agent-task/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/skills/leviuszen/meaning-assurance/agent-task"><img src="https://agentmods.dev/badge/skills/leviuszen/meaning-assurance/agent-task.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 34 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 770 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

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.

Model	Per session	Once invoked
Fable 5.1	$0.00034	$0.00770
Opus 5.5	$0.00014	$0.00308
Sonnet 5.5	$0.00007	$0.00154
Haiku 4.5	$0.00003	$0.00077

Measured 1mo ago against content hash db922a5a54ef, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

agent-task scanned grade A with 0 findings 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 1mo 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.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/agent-task/SKILL.md · 60 lines

How it starts

The opening of the file, as written. The whole thing — 60 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Task

Use this skill when the user wants Codex to delegate a bounded implementation, research, or exploration task to an external agent.

Rules

Codex remains the primary conversation and final reviewer.
Create a task packet before launching any external agent.
Keep the task narrow enough for one result file and one optional patch.
Do not paste or log secrets, API keys, launcher contents, or provider environment variables.
External agents should write outputs into the task folder.
Do not claim the delegated task is complete until Codex has collected and reviewed result files.

Workflow

Identify the workspace root and allowed paths.
Choose claude-code, reasonix, or manual.
Run New-AgentTask.ps1 with the task, context, mode, target agent, and workspace root.
For implementation work where the external agent may edit repository files, create an isolated worktree before launch. See Isolation For Implementation Tasks.
Launch the selected agent:

Formal non-interactive worker: Invoke-AgentTask.ps1 after isolated worktree creation. It dispatches to Claude Code or Reasonix from target_agent.
Claude Code direct compatibility runner: Invoke-ClaudeCodeTask.ps1
Reasonix direct compatibility runner: Invoke-ReasonixTask.ps1
Reasonix worker execution expects the reasonix CLI to be on PATH, or REASONIX_COMMAND / -ReasonixCommand to point to the real CLI shim.
Close Reasonix Desktop before a CLI worker run. The runner preflights the Desktop process and translates CLI session-lock failures into a clear retry instruction.
On Windows, the Reasonix runner denies Bash(*), confines file writers to the isolated worktree with workspace_root, restores any original reasonix.toml, and leaves shell verification to Codex after collection.
Treat reasonix-stdout.log and reasonix-stderr.log as redacted execution evidence. Canonical agent-result.md is cleaned and must start with # Agent Result.
- Claude Code manual session: Start-ClaudeTask.ps1
- Manual: give the task folder path to the user.

Wait for agent-result.md, result.md, or patch.diff.
Run Collect-AgentResult.ps1; when an isolated worktree exists, the collector prints a git status review section.
Review the output before applying any patch or recommendation.

Read the full file on GitHub · 60 lines

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.

1mo ago First seen · 60 lines · 34 tokens per session scan A db922a5a54ef

Subscribe to this mod's changes

agent-task is a skill published in the GitHub repository leviuszen/meaning-assurance (2 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 34 tokens to every session and 770 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Building agents Claude Code Multi-Agent Planning coding-agents human-in-the-loop local-first powershell

webgl-particle-galaxy

A real-time particle galaxy — tens of thousands of additive-blended GPU points spiraling around a bright core, their orbits solved entirely in the vertex shader (from glVertexID), so the whole cloud draws in one call. Rendered as a single self-contained index.html. Use when the brief asks for a "particle field"…

nexu-io/open-design · 126 tokens

webgl-liquid-metal

A real-time liquid-metal shader — a domain-warped noise field shaded as molten chrome, with a sweeping specular highlight over an iridescent thin-film (cosine-palette) sheen. No textures. Rendered as a single self-contained index.html. Use when the brief asks for "liquid metal", "molten chrome", "iridescent"…

nexu-io/open-design · 121 tokens

webgl-neon-grid

A real-time synthwave scene — a perspective floor grid scrolling toward a banded retro sun under a neon starfield, all from one fragment shader. No textures. Rendered as a single self-contained index.html. Use when the brief asks for "synthwave", "outrun", "retro 80s", "neon grid", "perspective grid", "vaporwave", or…

nexu-io/open-design · 115 tokens

agent-task