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 maps-geographygit 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/maps-geography)<a href="https://agentmods.dev/skills/asgeirtj/system_prompts_leaks/maps-geography"><img src="https://agentmods.dev/badge/skills/asgeirtj/system_prompts_leaks/maps-geography/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/maps-geography"><img src="https://agentmods.dev/badge/skills/asgeirtj/system_prompts_leaks/maps-geography.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- Snyk fail
- 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 | $0.00029 | $0.00800 |
| Opus 5 | $0.00015 | $0.00400 |
| Sonnet 5 | $0.00006 | $0.00160 |
| Haiku 4.5 | $0.00003 | $0.00080 |
Grade A, and why
maps-geography 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 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.
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.
Copies of this mod
1 near-identical copy found in the catalogue:
- maps-geography — 91% identical, 3 lines differ
What it actually says
Maps & geography
Geographic maps are data problems, not drawings: never freehand country outlines, coastlines, or street layouts — hand-drawn geography is reliably wrong, and users notice. Load real geometry and render it.
Build every map page as plain HTML — a .html file with ordinary tags, NEVER a .dc.html Design Component, even when every other design in the project is one: DC confines scripts to , whose mount timing races the map container — the same call the data-viz and 3D skills make.
For decks, docs, graphics, and animations — anything static or exported — render TopoJSON geometry with d3-geo: fetch https://cdn.jsdelivr.net/npm/[email protected]/countries-110m.json (Natural Earth data, public domain; the URL is version-pinned — use it exactly), convert with topojson.feature(topology, topology.objects.countries), and draw with d3.geoPath() under a projection chosen for the job (d3.geoNaturalEarth1 for the whole world; d3.geoMercator().fitSize(...) to zoom a region). d3-geo ships inside the d3 bundle below. Load the libraries ONLY through these exact pinned, hash-verified tags, in . These tags fail closed if tampered with; any other script you add would load unverified — so do not change versions, URLs, or hashes, and add nothing else from a CDN:
Inline SVG from d3 also exports cleanly to PNG and PDF, which live map tiles do not — so exported deliverables always get d3 geometry, never an embedded tile map.
For street-level interactive maps — prototypes, websites, anything the user pans and zooms — use Leaflet with OpenStreetMap tiles, loaded ONLY through these exact tags (the stylesheet is required: without leaflet.css the tiles render scrambled):
Create the map with L.map(...) and L.tileLayer('https://tile.openstreetmap.org/{z}/{x}/{y}.png', { attribution: '© OpenStreetMap contributors' }). The attribution string is OpenStreetMap's license requirement — never omit it.
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 · 26 lines · 29 tokens per session scan A 18e35f5715c5
maps-geography is a skill published in the GitHub repository asgeirtj/system_prompts_leaks (64,614 stars, last pushed yesterday), licensed CC0-1.0. It adds 29 tokens to every session and 800 once invoked, about $0.0001 per session on Opus 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 2026-08-30.
Other skills, from other repositories
neuroskill-bci
Use live BCI cognitive and mood state from NeuroSkill.
saelens
Train sparse autoencoders to interpret model features.
manim-video
Manim CE animations: 3Blue1Brown math/algo videos.
drug-discovery
Drug discovery: ChEMBL search, drug-likeness, interactions.
jupyter-notebook
Iterative Python via live Jupyter kernel (hamelnb).
bioinformatics
Gateway to 400+ genomics and computational biology skills.