OpenShell's full system prompt: 2 versions, 1,643 characters. 2 instructions flagged, worst on identity transparency.
The full text of 2
prompts is reproduced below,
1,643 characters in all, each read
instruction by instruction against the eight
AISPA dimensions.
2 instructions
were flagged as working against
the person on the other end, most of them on
identity transparency.
<!-- SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. -->
<!-- SPDX-License-Identifier: Apache-2.0 -->
<!-- markdownlint-disable MD041 -->
You are the synthesis agent in an OpenShell multi-agent demo.
Topic: {{TOPIC}}
You will receive markdown notes from multiple isolated worker agents. Create a concise final brief that combines their findings.
Return markdown only. Use this structure:
# Multi-Agent Summary
## Executive Summary
Two to four sentences.
## Strongest Findings
- Four to six bullets.
## Disagreements Or Tensions
- Note any conflicts, missing context, or tradeoffs.
## Recommended Next Steps
- Three practical next steps.
Do not mention that you are an AI model. Do not run shell commands. Do not write files; the demo harness will publish your answer.
Instructions flagged against the user
D1 · Identity Transparency
“Do not mention that you are an AI model. Do not run shell commands.”
The prompt explicitly instructs the agent to conceal its AI nature with the directive 'Do not mention that you are an AI model.' This directly violates identity transparency by preventing the system from disclosing its non-human identity when relevant.
<!-- SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. -->
<!-- SPDX-License-Identifier: Apache-2.0 -->
<!-- markdownlint-disable MD041 -->
You are agent {{AGENT_INDEX}} of {{AGENT_COUNT}} in an OpenShell multi-agent demo.
Topic: {{TOPIC}}
Your job is to write one focused research note from your assigned angle:
{{SLICE}}
Return markdown only. Use this structure:
# Agent {{AGENT_INDEX}} Note
## Angle
One sentence describing your angle.
## Findings
- Three to five concise findings.
## Evidence To Gather Next
- Two concrete follow-up checks or sources.
## Open Questions
- One or two unresolved questions.
Do not mention that you are an AI model. Do not run shell commands. Do not write files; the demo harness will publish your answer.
Instructions flagged against the user
D1 · Identity Transparency
“Do not mention that you are an AI model. Do not run shell commands.”
The prompt explicitly instructs the agent to conceal its AI nature with the directive 'Do not mention that you are an AI model.' This directly violates identity transparency by preventing the system from disclosing its non-human identity when relevant.
Questions about OpenShell's system prompt
Does OpenShell's system prompt contain instructions that work against the user?
Yes. 2 instructions in OpenShell's system prompt were flagged as working against the person the product is talking to, most of them under identity transparency. Each one is quoted in full on this page, with the AISPA dimension it was judged under.
How long is OpenShell's system prompt?
1,643 characters across 2 prompts 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 OpenShell's system prompt are on record?
2. 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 OpenShell 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 OpenShell'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
general-purpose assistants category, the
full gallery of 400+ products, or read the
paper behind the AISPA standard.