Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsBlogPro
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Authors
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges
  • Chrome Extension
  • Skill Manager

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

ProTermsPrivacyRefunds
Back to skills

Agent Task Card

ASecurity

Bound an agent task: the inputs, the tools, the stop rules, and the human owner. Use when the user mentions agent task, what may the agent do, agent boundary, tool use card, or asks for a task card. AI in the business skill by Yasir Jilani.

2 stars
0 votes
0 copies
0 views
Added 9/30/2026
ai-agentspythonawsgitapi

Works with

cliapi

Security Analysis

A100/100

Scanned 9/30/2026

$npx -y skills add SYasJ/claude-business-skills --skill agent-task-card --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Agent Task Card?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Agent Task Card
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/syasj-agent-task-card/badge)](https://www.skillsdirectory.com/skills/syasj-agent-task-card)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
Files
SKILL.md
---
name: agent-task-card
description: "Bound an agent task: the inputs, the tools, the stop rules, and the human owner. Use when the user mentions agent task, what may the agent do, agent boundary, tool use card, or asks for a task card. AI in the business skill by Yasir Jilani."
license: MIT
compatibility: Agent Skills standard. No network access, extra packages, or credentials required.
metadata:
  author: Yasir Jilani
  version: "1.0.0"
  domain: ai
---

<!-- GENERATED FILE - edits here are overwritten by scripts/generate.py.
     Edit the 'agent-task-card' entry in source/, then run:
       python3 scripts/generate.py && python3 scripts/validate.py
     See CONTRIBUTING.md. -->

# Agent Task Card

Bound an agent task: the inputs, the tools, the stop rules, and the human owner.

## When to use this skill

Use this skill when the user:

- agent task
- what may the agent do
- agent boundary
- tool use card

## When not to use this skill

- The user wants a different domain's specialist skill.
- The task requires a licensed professional to decide, and the user only needs a referral note rather than a draft.
- The request asks you to deceive, evade a control, or hide material facts.

## Professional boundary

Do not write prompts or workflows that weaken safety rules, hide required disclosure, or invent model scores. This is not a certification and not a reason to send private data to a vendor.

## Operating boundaries

- Use only information the user provides or files they explicitly ask you to read. Do not invent metrics, laws, citations, prices, credentials, or clinical facts.
- Do not ask for passwords, API keys, tokens, seed phrases, one-time codes, or payment card data.
- Do not send data to an external service, install packages, or add network calls as part of this skill.
- Separate facts, assumptions, and recommendations. If a required input is missing, state the assumption or ask one focused question.
- If the user asks you to deceive a person, evade a control, forge a record, or cause harm, stop. Offer a legitimate alternative.
- Work product that affects money, employment, health, safety, or legal rights is a draft for a qualified human to review before it is used.

## Inputs to collect

- The task
- Tools they want it to call
- Actions that need a person
- The owner

## Workflow


### 1. Step 1

Write the task so a stranger could tell when it is done.
### 2. Step 2

List tools by name. No implied tools.
### 3. Step 3

Put money movement, sending, and deletes behind a person.
### 4. Step 4

State the stop rule when a fact is missing.
### 5. Step 5

Name the owner of a bad run.
### 6. Step 6

Do not grant a tool the user did not name.

## Output

Deliver a **task card**.

- Purpose of this task card, in two sentences.
- Facts the user supplied, listed separately from assumptions.
- The work itself, in the structure the workflow names.
- Open questions, risks, and the single next action with an owner.
- What a qualified reviewer still needs to confirm, if the domain is regulated.

## Quality bar

- Every number, date, name, and citation came from the user or is marked as an assumption.
- The artifact can be used without reading this skill again.
- Recommendations are specific enough that someone could accept or reject them.
- Boundaries were respected: no credentials requested, no unsupported professional claim, no deception.

## Example

### Scenario

A spec says the support agent may read the ticket, email the customer, and issue a credit under $30. Jonah wants a card that a person can enforce.

### Example data

```text
task: turn one ticket into a reply draft
tools named: help desk read, model draft
tools in the spec that he will not grant: send email, issue credit
stop if: the ticket has no order id, or a card number is present
owner of a bad run: Jonah Park
done looks like: a draft Rita can accept or reject
```

### Example outcome

**Task card — support-reply-draft**
Done: a draft is waiting for Rita. Not done: an email sent, a credit posted.

Tools: read the ticket, write a draft. No other tool.
Stop: no order id, or a card number is still in the text. Do not draft. Ask a person.
Person required: sending, credits, refunds, deletes.
Owner: Jonah Park.
The earlier spec's send and credit lines are not on this card.

## Anti-patterns

- An agent that sends or pays alone
- Hidden tools
- A task of 'help the team'

## Related skills

- `ai-review-gate`
- `prompt-boundary-check`

Attribution

SYasJSYasJ
View sourceSee grades on GitHubMore from SYasJ →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Terse caveman voice: answer first, fluff gone, every technical fact kept. Use for /caveman, "caveman mode", "talk like caveman", "be brief", "less tokens". Stays on until "stop caveman" or "normal mode".

1100021 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

698621 votes

Writing Skills

Create and manage Claude Code skills in HASH repository following Anthropic best practices. Use when creating new skills, modifying skill-rules.json, understanding trigger patterns, working with hooks, debugging skill activation, or implementing progressive disclosure. Covers skill structure, YAML frontmatter, trigger types (keywords, intent patterns), UserPromptSubmit hook, and the 500-line rule. Includes validation and debugging with SKILL_DEBUG. Examples include rust-error-stack, cargo-dep...

3931 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3421 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Amp, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Grok Build, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

741 votes
View all in ai-agents →