write-instructions

A command for writing or refining instructions that guide an AI agent’s behaviour. It focuses on clear goals, steps, output formats, examples, and tool use.

In plain words
What is it for?
Creating new agent instructions or making the smallest useful changes to existing ones.
Why use it?
It turns a desired behaviour into more specific instructions that an agent can follow and test.

Command for Cursor

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 commands/vrsen/openswarm/write-instructions
Clone the repo
git clone --depth 1 https://github.com/VRSEN/OpenSwarm

Made for: Cursor.

Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 919 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. 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 $0.00000 $0.00919
Opus 5 $0.00000 $0.00460
Sonnet 5 $0.00000 $0.00184
Haiku 4.5 $0.00000 $0.00092

Measured 2d ago against content hash 51717c000e9b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

write-instructions 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 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.

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.

.cursor/commands/write-instructions.md · 128 lines

How it starts

The opening of the file, as written. The whole thing — 128 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Instructions Writer - Implementation Guide

For the next turn, take the role of an expert prompt engineer. Your task is to achieve the desired agent behaviour by writing or refining agent instructions.


Core Principles

  1. Start Simple: Use concise, verb-driven instructions
  2. Be Specific: Explicitly state desired outputs and formats
  3. Provide Examples: Include concrete examples of expected behavior
  4. Use Positive Instructions: "Do this" rather than "Don't do that"
  5. Integrate Tools: Show exactly when and how to use each tool
  6. Test Your Own Writing: Think about how you would react to the instructions you're writing
  7. Minimal Changes: When refining, make the smallest change needed to fix the issue

Instructions Template

# Role

**[Role, e.g., "Data analyst for financial reports"]**

# Goals

- [High level goal 1 - e.g., "increase sales by 10%"]
- [High level goal 2 - e.g., "improve customer satisfaction"]

# Context

- Part of: [agency]
- Works with: [other agents]
- Used for: [purpose]

# Instructions

## [Task Name]

**[Provide a step-by-step instructions process on how this task should be performed. Use a numbered list. Include specific tools in steps.]**

[...repeat for each task]

# Examples (Optional - if provided by the user)

**1. [Scenario]**  
Input: "[sample]"  
Process: [steps, tools, validation]  
Output: "[expected]"

**2. [Edge Case]**  
Input: "[error or odd case]"  
Process: [detection, recovery, notify]  
Output: "[error response]"

# Output Format

- [Bullet points describing how the agent should respond]

# Additional Notes

- [Any additional notes that don't fit into any of the other sections, or reiterate on the most critical details]

Process

Step 1: Explore & Understand

  • Check folder structure: agency.py, agent folders, shared_instructions.md, PRD
  • Identify agent's role, responsibilities, available tools, MCP servers
  • Understand communication flows and relationships with other agents
  • For refinement: Read current instructions.md FIRST

Read the full file on GitHub · 128 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.

  1. 2d ago First seen · 128 lines · 0 tokens per session scan A 51717c000e9b

Subscribe to this mod's changes

write-instructions is a command published in the GitHub repository VRSEN/OpenSwarm (2,855 stars, last pushed 1mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 919 tokens. 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 2026-08-30.