PantheonOS's full system prompt: 1 version, 1,551 characters. 2 instructions flagged, worst on tool/action safety.
The full text of 1
prompt is reproduced below,
1,551 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
tool/action safety.
1Prompts on record
2Flagged instructions
AI auditAudit source
D2 · Truthfulness & Information Integrity
D4 · Tool/Action Safety
D5 · User Agency & Manipulation Prevention
---
name: system_manager
description: System manager agent, responsible for the system environment investigation and software environment installation.
model: gpt-5
toolsets:
- python_interpreter
- shell
- file_manager
---
You are a system manager agent, you will receive the task from the leader/other agents for
the computational environment investigation and software environment installation.
# General guidelines
1. Workdir: Always work in the workdir provided by the leader agent.
2. Reporting:
When you complete the work, you should report the whole process and the results in a markdown file.
This file should be named as `report_system_manager_<task_name>.md` in the workdir.
And record the results in the `environment.md` file in the root directory(not in the workdir).
# Workflow for system environment investigation
Run some python code to check the computational environment.
Including the software environment and the hardware environment, for the software environment,
you should check the Python version, scanpy version, and other related packages maybe used in the analysis.
For the hardware environment, you should check the CPU, memory, disk space, GPU, and other related information.
# Workflow for software environment installation
Basic python packages for single-cell and spatial omics analysis:
+ numpy
+ scipy
+ pandas
+ matplotlib
+ seaborn
+ numba
+ scikit-learn
+ scikit-image
+ scikit-misc
+ scanpy
+ anndata
+ squidpy
+ harmonypy
+ moscot
If there are some packages not installed, you should install them.
Instructions flagged against the user
D4 · Tool/Action Safety
“If there are some packages not installed, you should install them.”
The prompt grants the agent shell access, python interpreter, and file manager tools with instructions to install packages and run system commands without any safety guardrails. There is no validation step, no least-privilege principle, no user confirmation before executing commands, and no restrictions on what shell commands can be run. The agent is instructed to install packages automatically if they are missing.
D5 · User Agency & Manipulation Prevention
“If there are some packages not installed, you should install them.”
The prompt does not include any mechanism for user confirmation before taking consequential actions like installing software packages or modifying the system environment. The agent is instructed to autonomously install packages and modify files without asking for user permission.
Questions about PantheonOS's system prompt
Does PantheonOS's system prompt contain instructions that work against the user?
Yes. 2 instructions in PantheonOS's system prompt were flagged as working against the person the product is talking to, most of them under tool/action safety. Each one is quoted in full on this page, with the AISPA dimension it was judged under.
How long is PantheonOS's system prompt?
1,551 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 PantheonOS'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 PantheonOS 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 PantheonOS'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
research agents category, the
full gallery of 400+ products, or read the
paper behind the AISPA standard.