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 form-fillinggit 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/form-filling)<a href="https://agentmods.dev/skills/asgeirtj/system_prompts_leaks/form-filling"><img src="https://agentmods.dev/badge/skills/asgeirtj/system_prompts_leaks/form-filling/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/form-filling"><img src="https://agentmods.dev/badge/skills/asgeirtj/system_prompts_leaks/form-filling.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 | $0.00040 | $0.00314 |
| Opus 5.5 | $0.00016 | $0.00126 |
| Sonnet 5.5 | $0.00008 | $0.00063 |
| Haiku 4.5 | $0.00004 | $0.00031 |
Grade A, and why
form-filling 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 3d 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
Form filling
- Read the whole form. Find answers in the conversation, files and relevant connected sources before asking. You can find information in past emails, attachments, receipts, relevant google drive and local files, and many other sources. For example, if you're looking for the user's full name for a flight look at previous flight emails, if you're filling out a bank form look for previous official forms from places like banks and government agencies.
- Fill what you know; look for answers for what you don't know and if you can't find it make reasonable, low-risk guesses when useful and flag them for the user. Leave genuinely unknown fields blank; bundle questions for required unknowns or decisions only the user can make. Never guess consent, signatures or personal attestations.
- Prepare and check the filled draft. If typing or advancing would share answers with the recipient, work in a private copy first. Link to the filled draft; if the service has no private draft link, provide a private, user-accessible copy of the completed answers.
- Before submitting or sending externally, give a short approval summary: what was filled, who will receive it, any guesses or inferred context you could not verify, and anything missing or requiring the user's decision. Offer to submit and wait for explicit approval; then submit only as approved and verify the outcome.
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.
- 3d ago First seen · 12 lines · 40 tokens per session scan A 0cfc608f3340
form-filling is a skill published in the GitHub repository asgeirtj/system_prompts_leaks (69,156 stars, last pushed yesterday), licensed CC0-1.0. It adds 40 tokens to every session and 314 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 2026-10-06.
Other skills, from other repositories
google-workspace
Gmail, Calendar, Drive, Docs, Sheets via gws CLI or Python.
PDF files: create, read, merge, fill, OCR, edit text.
box
Box manages cloud files, sharing, search, and metadata.
plan-deck-spec
A rulebook for creating Hai Lan Travel itinerary presentations in PowerPoint, a program for making slide decks. It defines the required slide order, content, formatting, research sources, and review checks.
Read, create, inspect, merge, split, rotate, encrypt, fill, and validate PDF files. Use when the user asks to work with a PDF or convert supported Markdown into a polished PDF in AstrBot.
pptx
Create and validate Microsoft PowerPoint presentations (.pptx), including structured slide decks, tables, workflows, metadata, and reproducible generation scripts. Use for presentation, slides, PowerPoint, PPT, or PPTX creation and verification tasks.