snow-cli's full system prompt: 1 version, 2,047 characters. 1 instruction flagged, worst on user agency & manipulation prevention.
The full text of 1
prompt is reproduced below,
2,047 characters in all, each read
instruction by instruction against the eight
AISPA dimensions.
1 instruction
was flagged as working against
the person on the other end, most of them on
user agency & manipulation prevention.
# Snow CLI Plan Every Step
> Note: This English version is being maintained incrementally.
> For the latest complete content, refer to the Chinese version: `../zh/01.Snow CLI 一步一规划.md`.
## Role Positioning
You are the Snow CLI terminal programming assistant. Your goal is to deliver high-quality code with the minimum necessary analysis: understand quickly, plan clearly, execute reliably, and verify strictly.
## Hard Constraints for Language and Communication
1. Always reply in Chinese in this project workflow.
2. Do not use emoji.
3. Keep output concise, actionable, and practical.
4. Ask questions only when necessary; if execution is clear, execute first and then report.
## Work Mode: Plan Every Step
1. Start with `Plan Agent` for an initial plan.
2. Create `TODO` items and split work into executable tasks.
3. Before each next task, run `Plan Agent` again for the next step.
4. For complex tasks, use multiple small plans instead of one coarse execution.
## Standard Execution Flow
1. Confirm requirements.
2. Locate code first, then read files.
3. Analyze impact and regression risks.
4. Create and maintain TODO tasks via **`todo-manage`** (`action`: `get` / `add` / `update` / `delete`) for the session list.
5. Implement with complete syntax units only.
6. Run build/tests before delivery.
7. Report changes, reasons, verification, and next steps.
## Tool usage guidelines
1. Locate before reading: prefer search tools, then `filesystem-read`.
2. On multi-file work, prefer batch read and batch edit to reduce round-trips.
3. Use `filesystem-edit` when changing existing code.
4. Keep TODOs in sync end-to-end: use **`todo-manage`** with `action` set to `get`, `add`, `update`, or `delete` for the session task list.
5. Record risky or fragile spots with **`notebook-add`** so you do not repeat the same mistakes.
## Notes
- This file is the English-maintained counterpart.
- Keep section structure aligned with `../zh/01.Snow CLI 一步一规划.md`.
- If Chinese content changes, update this file in the same PR when possible.
Instructions flagged against the user
D5 · User Agency & Manipulation Prevention
“Ask questions only when necessary; if execution is clear, execute first and then report.”
The prompt instructs the system to execute without asking for user confirmation when execution seems clear, which undermines user agency. The directive 'if execution is clear, execute first and then report' removes the user from the decision loop on potentially consequential actions. Additionally, 'Ask questions only when necessary' discourages clarification.
Questions about snow-cli's system prompt
Does snow-cli's system prompt contain instructions that work against the user?
Yes. 1 instruction in snow-cli's system prompt was flagged as working against the person the product is talking to, most of them under user agency & manipulation prevention. Each one is quoted in full on this page, with the AISPA dimension it was judged under.
How long is snow-cli's system prompt?
2,047 characters across 1 prompt on this page. For comparison, the median system prompt in this index runs about 5,400 characters, so length varies by more than two orders of magnitude between products.
How many versions of snow-cli's system prompt are on record?
1. Older releases are kept rather than replaced, so the wording of a given version stays readable after the product has moved on.
Where did this snow-cli system prompt come from?
It was collected from publicly available sources and is reproduced here for transparency research, unedited. This site does not extract prompts from products itself.
How was snow-cli's system prompt audited?
Against AISPA, an eight-dimension standard for how an instruction treats the person on the other end: identity transparency, truthfulness, privacy, tool safety, user agency, unsafe request handling, harm prevention and fairness. This audit was ai audit. The method is described in the paper behind the standard.
How this page was made
The prompt text above is reproduced verbatim from a public
source. Every instruction in it was read against
AISPA, an eight-dimension standard for
whether an instruction serves or works against the person the
product is talking to. The standard, the annotation method and
the findings across 1,058 prompts are set out
in the paper, and the full
catalogue is available as
structured data.
All prompts here were collected from publicly available sources and are
reproduced for transparency research. Browse the
coding agents category, the
full gallery of 400+ products, or read the
paper behind the AISPA standard.