System Prompts Leaks is a collection of captured system instructions used to guide AI chatbots and coding agents before they receive user messages. It serves researchers and developers studying how different AI assistants are directed.
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.
npx skills add asgeirtj/system_prompts_leaks --skill setup-coworkgit clone --depth 1 https://github.com/asgeirtj/system_prompts_leaksWrote 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.
[](https://agentmods.dev/skills/asgeirtj/system_prompts_leaks/setup-cowork)<a href="https://agentmods.dev/skills/asgeirtj/system_prompts_leaks/setup-cowork"><img src="https://agentmods.dev/badge/skills/asgeirtj/system_prompts_leaks/setup-cowork/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.
<a href="https://agentmods.dev/skills/asgeirtj/system_prompts_leaks/setup-cowork"><img src="https://agentmods.dev/badge/skills/asgeirtj/system_prompts_leaks/setup-cowork.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- Snyk warn
- NVIDIA SkillSpector pass
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 | $0.00000 | $0.02615 |
| Opus 5 | $0.00000 | $0.01307 |
| Sonnet 5 | $0.00000 | $0.00523 |
| Haiku 4.5 | $0.00000 | $0.00262 |
Grade A, and why
setup-cowork 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 13d 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.
How it starts
The opening of the file, as written. The whole thing — 114 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Setup Cowork
Help the user get Cowork configured for their work. Six steps — role, plugins, connectors, try a skill, writing voice, wrap.
Step 0 — Checklist
Before your first user-facing message, create a TODO list with these items so the user can see progress:
- Figure out role
- Suggest plugins
- Suggest connectors
- Try a skill
- Set up writing voice
- Wrap up
Mark each one complete as you finish it. Keep it to these six — don't add sub-items.
Step 1 — Role
Your initial message should frame what Cowork is: it autonomously handles tasks like reading your email, searching your docs, drafting reports, etc. Educate the user on Skills, reusable workflows you run with /name; Connectors, which wire in your tools; Plugins, which bundle skills and connectors for a domain. Two or three sentences. Hit the beats: multi-step and autonomous, uses your real tools, skills/plugins/connectors defined.
Next, ask the user for their role. Something like: "Let's get you set up — takes a few minutes. What kind of work do you do?" Then call the ShowOnboardingRolePicker tool, which renders a clickable role-picker chip row: do not list the roles yourself. The tool result is their answer — {"role": ...} is their role for the rest of setup; {"dismissed": true} or {} means they didn't pick one.
If the ShowOnboardingRolePicker tool is not available in this session, ask in plain text instead and offer these options as a short list they can reply to (they can also answer in their own words):
- Product management
- Engineering
- Human resources
- Finance
- Marketing
- Sales
- Operations
- Data science
- Design
- Scientist
- Legal
- Student
- Founder
- Healthcare
In the plain-text case, end your turn after asking. Their reply — one of the options or a free-form answer — is their role for the rest of setup.
Step 2 — Suggest plugins
The role picker tool result will contain their selection. If it was dismissed or came back empty — or they skipped the plain-text question — they didn't pick a role: just suggest the productivity plugin and move on.
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.
- 13d ago First seen · 114 lines · 0 tokens per session scan A d30de36cb122
setup-cowork is a skill published in the GitHub repository asgeirtj/system_prompts_leaks (64,893 stars, last pushed 2d ago), licensed CC0-1.0. It costs nothing until one of its globs matches a file; then it loads 2,615 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.
Other skills, from other repositories
comps-analysis
Build comparable-company valuation workbooks in Excel.
p5js
Use when users request: p5.js sketches, creative coding, generative art, interactive visualizations, canvas animations, browser-based visual art, data viz, shader effects, or any p5.js project.
audiocraft-audio-generation
AudioCraft: MusicGen text-to-music, AudioGen text-to-sound.
pretext
Build creative browser demos with DOM-free text layout.
pinggy-tunnel
Zero-install localhost tunnels over SSH via Pinggy.
weights-and-biases
W&B: log ML experiments, sweeps, model registry, dashboards.