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 vercelgit 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/vercel)<a href="https://agentmods.dev/skills/asgeirtj/system_prompts_leaks/vercel"><img src="https://agentmods.dev/badge/skills/asgeirtj/system_prompts_leaks/vercel/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/vercel"><img src="https://agentmods.dev/badge/skills/asgeirtj/system_prompts_leaks/vercel.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 | $0.00052 | $0.00512 |
| Opus 5.5 | $0.00021 | $0.00205 |
| Sonnet 5.5 | $0.00010 | $0.00102 |
| Haiku 4.5 | $0.00005 | $0.00051 |
Grade A, and why
vercel 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 14d 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.
What it actually says
Vercel
Use the installed vercel CLI. Start with vercel status. If it reports
not_connected, run vercel authorize-url and share only the returned
connect_url with the user.
OAuth uses Dynamic Client Registration and PKCE through authd. Vercel declares the client public and issues no client secret, so no shared credential enters Muse and credentials must never be requested in chat. Vercel exposes only identity and session OAuth scopes for this MCP server, so read-only-by-default behavior is enforced by the connector permissions below rather than a narrower provider scope.
Run vercel list-tools to inspect the live provider catalogue and schemas,
then call an advertised tool with:
vercel call-tool --name <tool> --arguments-json '<json-object>'
The reviewed catalogue follows Vercel's current public MCP tool documentation,
including create_deployment. The former deploy_to_vercel name and other
recently replaced names remain accepted only when the live server still
advertises them during rollout. Always use the live schema returned by
list-tools; for example, current create_deployment arguments place the
deployment definition under requestBody.
list-tools exposes only reviewed Vercel tools and includes each tool's
hatch_permission, hatch_action, and hatch_permission_label. Unknown or
new provider tools remain unavailable until reviewed. Ordinary reads follow
the user's connector settings. Decrypted secrets, deployment or sandbox file
contents, deployments, purchases, credential creation, security changes,
sandbox execution, and other mutations use separate granular permissions that
ask by default. Vercel advertises identity/session OAuth scopes only, so there
is no provider-side incremental scope to request for an individual tool.
Purchase tools can create immediate, non-refundable charges. Obtain a quote when the live tool contract provides one, show it to the user, and never infer confirmation. Do not retry any failed or timed-out write automatically because its side effect may have completed.
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.
- 14d ago First seen · 52 lines · 52 tokens per session scan A a4fad1284a68
vercel is a skill published in the GitHub repository asgeirtj/system_prompts_leaks (69,863 stars, last pushed 3d ago), licensed CC0-1.0. It adds 52 tokens to every session and 512 once invoked, about $0.0002 per session on Opus 5.5. 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 2026-10-05.
Other skills, from other repositories
anatomy
Explain an idea as an interactive isometric machine figure.
ascii-video
ASCII video: convert video/audio to colored ASCII MP4/GIF.
creative-ideation
Generate ideas via named methods from creative practice.
llava
Vision-language chat: VQA, captioning, image dialogue.
maps
Geocode, POIs, routes, timezones via OpenStreetMap/OSRM.
deployment-patterns
Deployment workflows, CI/CD pipeline patterns, Docker containerization, health checks, rollback strategies, and production readiness checklists for web applications. Use when setting up deployment infrastructure or planning releases.