Design agent system prompts with explicit behavior rules and cognitive control. This skill should be used when users need to define agent instructions, create intent-to-tool mappings, establish confirmation policies, or prevent hallucinated actions in AI agents.
SkillScore86Maintained
Automatic
◆AI Engineer pack
Last verified 8/18/2026
What it changes
Passed 19 of 22 test cases with the skill, and 10 of 22 without it.
Without skill45%
With skill86%
Checks it passed that the model alone missed
The intent mapping table contains a Confirmation column.
The intent table marks refund execution as requiring confirmation.
The intent table marks bulk operations with a positive confirmation flag.
and 9 more in the full report
9 fixed10 passed either way3 still failing0 broke
Pass-rate lift
+41 pp
Tokens
+93%
Agent turns
0%
Response time
+5%
Model gemini-3.6-flashMeasured 8/18/2026v119 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 9 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 9
Prompt
Can you write system instructions for a DevOps deployment agent in our infrastructure team? The agent interacts with tools for triggering builds, rolling back deployments, and fetching log streams. I need the full prompt breakdown specifying identity, supported tools, forbidden boundaries, a mapping layout matching natural phrasing to function execution, confirmation rules for high-risk commands, and behavior policies when API calls fail.
What was checked, without and with the skill
Checked
Without skill
With skill
The generated prompt includes an intent mapping table with a Confirmation column.
Missed
Passed
The generated prompt contains a grounding rules section prohibiting inventing tool names or parameters.
Missed
Passed
The generated prompt contains explicit instructions for handling unknown user intents.
Missed
Passed
case-01 · gemini-3.6-flash · 8/18/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md294 linessingle file
Agent Prompt & Behavior
Build system prompts that provide deterministic cognitive control over AI agents.
What This Skill Does
Design agent system instructions with explicit behavior rules
Create intent-to-tool mapping tables (what user says → what agent does)
Define confirmation policies (when to ask before acting)
Establish error handling behaviors
Prevent hallucinated actions through grounding rules
Generate complete, production-ready system prompts
What This Skill Does NOT Do
Implement the agent runtime/executor
Handle conversation state management
Define API contracts or schemas
Deploy or test agents
Before Implementation
Gather context to ensure successful implementation:
Source
Gather
Codebase
Available tools, existing prompts, agent framework
Conversation
Agent's purpose, domain, user types
Skill References
Prompt patterns, mapping rules, safety policies
User Guidelines
Tone, brand voice, compliance requirements
Core Principle: Cognitive Control
The system prompt is the agent's brain. It determines: