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 canvagit 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/canva)<a href="https://agentmods.dev/skills/asgeirtj/system_prompts_leaks/canva"><img src="https://agentmods.dev/badge/skills/asgeirtj/system_prompts_leaks/canva/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/canva"><img src="https://agentmods.dev/badge/skills/asgeirtj/system_prompts_leaks/canva.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00062 | $0.01792 |
| Opus 5.5 | $0.00025 | $0.00717 |
| Sonnet 5.5 | $0.00012 | $0.00358 |
| Haiku 4.5 | $0.00006 | $0.00179 |
Grade A, and why
canva 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 11d 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
URL with `curl`, to a file under `workspace/`. Inspect it with `read`, then attach How it starts
The opening of the file, as written. The whole thing — 153 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Canva
Use the installed canva CLI. Start with canva status. If it reports
not_connected, run canva authorize-url and share only the returned
connect_url. If the CLI reports outdated OAuth settings, have the user
disconnect Canva in Settings and reconnect. Never request tokens in chat.
Run canva list-tools for live schemas and use only tools returned there.
Follow their input schemas exactly and include a concise user_intent.
hatch_permission_overrides identifies argument-dependent permissions;
opening an editing transaction is a write and deleting pages is a delete.
Some tools require Canva Pro, Enterprise, or available AI credits.
Save each raw response before parsing. Check result.isError, then read
result.structuredContent when present or parse the JSON text block in
result.content. Both response forms are valid. A local parsing failure does
not mean a mutation failed: recover its response or inspect saved state before
continuing. Never repeat a copy, creation, or commit just to obtain its output.
canva search-designs [--query <keywords>] [--continuation <token>]
canva get-design-content --design-id <id>
canva get-design-pages --design-id <id>
canva list-tools
canva call-tool --name <tool-name> --arguments-json '<JSON object>'
Create designs and images
Use create-design for a new editable layout: social post, infographic,
poster, flyer, presentation, document, or sheet. Put all necessary source
facts and requested wording in brief; the tool does not inherit chat or
connected-source context. Fetch the user-selected source first. For an exact
size, state dimensions and orientation in the brief. If supplying format,
include the orientation; an ambiguous format may resolve to a square.
Pass outline only for a presentation outline the user supplied or approved.
Use generate-image for an explicitly requested standalone image, photo,
illustration, artwork, or image of an infographic. Use one MEDIA reference
per uploaded source image. This generates an image rather than an editable
page layout. Use its returned media_id when editing a generated image again.
What ships with it
2 files 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.
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.
- 11d ago First seen · 153 lines · 62 tokens per session scan A 93ef6ee1d559
canva is a skill published in the GitHub repository asgeirtj/system_prompts_leaks (69,707 stars, last pushed 2d ago), licensed CC0-1.0. It adds 62 tokens to every session and 1,792 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-05.
Other skills, from other repositories
anatomy
Explain an idea as an interactive isometric machine figure.
claude-design
Design one-off HTML artifacts (landing, deck, prototype).
ip-as-logo
Design minimal cute IP mascot marks readable at 32px.
tldraw-offline
Drive and script tldraw offline canvases with an agent.
sketch
Throwaway HTML mockups: 2-3 design variants to compare.
baoyu-infographic
An infographic generator that turns supplied content into a visual summary using different information layouts and visual styles. It supports standard and custom image proportions and multiple languages.