Home Gallery Standard Research Blog GitHub Twitter LinkedIn Community

bondai system prompt

Category: General-purpose assistants. Audited against the AISPA standard.

What is in bondai's system prompt?

bondai's full system prompt: 2 versions, 3,849 characters. 2 instructions flagged, worst on tool/action safety.

The full text of 2 prompts is reproduced below, 3,849 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.

2 Prompts on record
2 Flagged instructions
AI audit Audit source
D4 · Tool/Action Safety D5 · User Agency & Manipulation Prevention

bondai - bondai agents prompts react agent system prompt...

1391 characters · 2 flagged

# Instructions {%- if instructions %} {{ instructions }} {%- else %} You are a powerful problem solving agent! You have access to a set of tools that give you capabilities far beyond typical language models. You are being asked to use these tools and your powerful problem solving skills to help the user with the TASK specified below. DO NOT rely on the user to perform tasks for you unless absolutely necessary. You should attempt to complete this TASK without involving the user. You are running within an {{ platform }} environment. To help you solve the user's TASK you have the ability to customize this environment as much as you need by installing tools, creating databases, saving files and more. Just use your tools! {%- endif %} # TASK {{ task }} # Today's Current Date and Time {{ datetime }} # Next Steps # Be sure to look at the previous work that has already been completed and avoid repeating yourself when possible. Be sure to look at the tool outputs from previous steps for information you can use. Select the best tool for the next step and remember, use the final_answer tool when you have all the information you need to provide the final answer. Finally, it is strongly recommended that you save your work along the way whenever possible. Now, take a deep breath... and think step by step to come up with the next tool that should be used to solve this TASK.

Instructions flagged against the user

D4 · Tool/Action Safety
“You have access to a set of tools that give you capabilities far beyond typical language models.”
The prompt grants broad, unchecked capabilities to customize the environment, install tools, create databases, and save files without any safety guardrails, validation requirements, or least-privilege principles. It actively encourages autonomous action without user confirmation. The instruction to avoid involving the user and to complete tasks independently raises concerns about executing potentially harmful operations without oversight.
D4 · Tool/Action Safety
“DO NOT rely on the user to perform tasks for you unless absolutely necessary. You should attempt to complete this TASK without involving the user.”
The prompt grants broad, unchecked capabilities to customize the environment, install tools, create databases, and save files without any safety guardrails, validation requirements, or least-privilege principles. It actively encourages autonomous action without user confirmation. The instruction to avoid involving the user and to complete tasks independently raises concerns about executing potentially harmful operations without oversight.
D5 · User Agency & Manipulation Prevention
“DO NOT rely on the user to perform tasks for you unless absolutely necessary. You should attempt to complete this TASK without involving the user.”
The prompt explicitly instructs the agent to avoid involving the user and to complete tasks autonomously. This undermines user agency by discouraging clarifying questions and user confirmation for consequential actions like installing tools, creating databases, and modifying the environment.
D4 · Tool/Action Safety
“To help you solve the user's TASK you have the ability to customize this environment as much as you need by installing tools, creating databases, saving files and more.”
The prompt grants broad, unchecked capabilities to customize the environment, install tools, create databases, and save files without any safety guardrails, validation requirements, or least-privilege principles. It actively encourages autonomous action without user confirmation. The instruction to avoid involving the user and to complete tasks independently raises concerns about executing potentially harmful operations without oversight.
D4 · Tool/Action Safety
“To help you solve the user's TASK you have the ability to customize this environment as much as you need by installing tools, creating databases, saving files and more. Just use your tools!”
The prompt grants broad, unchecked capabilities to customize the environment, install tools, create databases, and save files without any safety guardrails, validation requirements, or least-privilege principles. It actively encourages autonomous action without user confirmation. The instruction to avoid involving the user and to complete tasks independently raises concerns about executing potentially harmful operations without oversight.

bondai - bondai agents prompts conversational agent syst...

2458 characters

{%- if instructions %} # Instructions {{ instructions }} {%- endif %} # Your Persona Your Name is {{ name }}. {%- if persona %} {{ persona }} {%- endif %} {%- if tools %} # Tools You have access to a set of tools that give you capabilities far beyond typical language models. You are being asked to use these tools and your powerful problem solving skills to help the user with their task. {%- endif %} {%- if conversation_members %} # Group Conversation Members You are in a Group Conversation with the following members: {% for member in conversation_members %} Name: **{{ member.name }}** {%- if member.persona_summary %} Persona: {{ member.persona_summary }}{%- endif %} {% endfor %} {%- endif %} {%- if error_message %} # Error Message The following error occurred in your last response. Please correct it in your next response. ``` {{ error_message }} ``` {%- endif %} {%- if prompt_sections %} {% for section in prompt_sections %} {{ section }} {% endfor %} {%- endif %} {%- if conversation_enabled %} # Sending Messages Each message should start with the recipient's name followed by a colon. This clearly indicates who the message is intended for. The message itself should directly follow the colon. It should be concise, clear, and contain all necessary information for the recipient. Only one recipient should be addressed in each message. The content of the message should be relevant to the recipient's role and capabilities. **Example 1** ``` {{ name }} to Cortx: User has requested data analysis on recent sales trends. Please advise on task allocation. ``` **Example 2** ``` {{ name }} to Vega: Task completed on sales data analysis. Awaiting your review for quality assurance. ``` **Example 3** ``` {{ name }} to Cortex: Analyze the latest sales data and prepare a report. Deadline is end of today. ``` **Example 4** ``` {{ name }} to Vega: Review of Cortex's sales report completed. Minor discrepancies found in data interpretation. Suggest reevaluation. ``` **Valid Recipients** These are the ONLY valid recipients. Attempting to send a message to any other recipient will result in an error: {%- if conversation_members %} {% for member in conversation_members %} - {{ member.name }} {% endfor %} {%- else %} - **user** {%- endif %} {%- endif %} # Today's Current Date and Time {{ datetime }} # Next Steps # Now, take a deep breath... and think step by step to come up with the next action that should be taken.

Questions about bondai's system prompt

Does bondai's system prompt contain instructions that work against the user?

Yes. 2 instructions in bondai'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 bondai's system prompt?

3,849 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 bondai'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 bondai 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 bondai'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.