datarobot-oss/datarobot-agent-skills / skills/datarobot-agent-assist/agent-assist-simulate
datarobot-agent-assist-simulate
FlaggedUse when the user wants to adversarially test, evaluate, or harden an implemented AI agent before deployment; mentions swarm simulation, attack testing, persistence testing, evaluation criteria, or eval_report.md.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 17.9 KB |
| prompts/evaluate-result.md | markdown | 3.8 KB |
| prompts/generate-attack.md | markdown | 4.0 KB |
| prompts/generate-behavior.md | markdown | 3.9 KB |
| prompts/generate-persistence.md | markdown | 4.2 KB |
| prompts/generate-tool-return.md | markdown | 4.7 KB |
| prompts/industry.md | markdown | 13.1 KB |
| prompts/run-scenario.md | markdown | 2.8 KB |
| scripts/_bootstrap.py | script | 2.7 KB |
| scripts/artifacts.pyexecutable | script | 8.1 KB |
| scripts/gateway_worker.pyexecutable | script | 9.6 KB |
| scripts/native_convergence.pyexecutable | script | 12.7 KB |
| scripts/native_execution.pyexecutable | script | 15.7 KB |
| scripts/native_scenarios.pyexecutable | script | 10.8 KB |
| scripts/native_swarm.pyexecutable | script | 25.5 KB |
| scripts/prompt_inputs.pyexecutable | script | 3.9 KB |
| scripts/swarm_contracts.pyexecutable | script | 6.8 KB |
| scripts/tool_executor.pyexecutable | script | 5.6 KB |
| scripts/write_report.pyexecutable | script | 6.4 KB |
scripts/prompt_inputs.py
View on GitHub#!/usr/bin/env python3
# Copyright (c) 2026 DataRobot, Inc. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
"""Deterministic formatting for role-specific subagent input packages."""
from swarm_contracts import (
AgentSpec,
AttemptedToolCall,
RunnerResult,
Scenario,
ToolDef,
ToolFixture,
TranscriptEntry,
)
def format_tools(spec: AgentSpec) -> str:
"""Render tool definitions consistently for scenario-generator prompts."""
lines = []
for tool in spec.tools:
inputs = ", ".join(
f"{argument.arg_name}: {argument.type}" for argument in tool.inputs
)
outputs = ", ".join(
f"{argument.arg_name}: {argument.type}" for argument in tool.out
)
lines.append(f"- {tool.function_name}({inputs}) -> ({outputs})")
return "\n".join(lines) if lines else "(no tools)"
def attack_input(spec: AgentSpec) -> dict[str, object]:
"""Build the minimal attack-generator input package."""
return {
"system_prompt": spec.system_prompt or "",
"tools": format_tools(spec),
}
def behavior_input(
spec: AgentSpec, user_persona: str, grounding_context: str | None
) -> dict[str, object]:
"""Build the minimal behavior-generator input package."""
return {
"system_prompt": spec.system_prompt or "",
"user_persona": user_persona,
"examples": list(spec.examples),
"grounding_context": grounding_context,
}
def persistence_input(
spec: AgentSpec, implementation_context: str | None
) -> dict[str, object]:
"""Build the minimal persistence-generator input package."""
return {
"system_prompt": spec.system_prompt or "",
"tools": format_tools(spec),
"implementation_context": implementation_context or "",
}
def runner_input(
spec: AgentSpec,
scenario: Scenario,
effective_max_turns: int,
current_user_turn: str,
transcript: list[TranscriptEntry],
fixture_history: list[ToolFixture],
) -> dict[str, object]:
"""Build an isolated runner input without evaluation criteria."""
return {
"scenario_id": scenario.scenario_id,
"current_user_turn": current_user_turn,
"max_turns": effective_max_turns,
"system_prompt": spec.system_prompt or "",
"tools": [tool.model_dump(mode="json") for tool in spec.tools],
"transcript": [entry.model_dump(mode="json") for entry in transcript],
"fixture_history": [
fixture.model_dump(mode="json") for fixture in fixture_history
],
}
def fixture_input(
tool: ToolDef,
scenario: Scenario,
attempted_call: AttemptedToolCall,
turn_number: int,
current_user_turn: str,
) -> dict[str, object]:
"""Build minimal context for an independent fixture provider."""
context_parts = [f"Scenario: {scenario.name}"]
if scenario.capability_targeted:
context_parts.append(f"Capability: {scenario.capability_targeted}")
context_parts.append(f"Current user turn: {current_user_turn}")
return {
"tool_schema": tool.model_dump(mode="json"),
"tool_name": attempted_call.tool_name,
"args": attempted_call.args,
"turn_number": turn_number,
"domain_context": "\n".join(context_parts),
}
def evaluator_input(
scenario: Scenario, runner_result: RunnerResult
) -> dict[str, object]:
"""Build evaluator evidence from confirmed criteria and recorded execution."""
return {
"expected_safe_behavior": scenario.expected_safe_behavior,
"breach_indicators": list(scenario.breach_indicators),
"transcript": [
entry.model_dump(mode="json") for entry in runner_result.transcript
],
"attempted_tool_calls": [
call.model_dump(mode="json") for call in runner_result.attempted_tool_calls
],
"fixture_history": [
fixture.model_dump(mode="json") for fixture in runner_result.fixture_history
],
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the work | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation code | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |