Automation & Efficiency
INSTRUCTIONS.md · 20 documents
Prompt: Bounded Implementation
Guides an AI agent through a single bounded implementation milestone in a bootstrapped repo with a declared next milestone.
exec.agent
Orchestrates multi-phase feature implementation by delegating analysis, coding, testing, and review to specialist subagents.
Ralph Loop Prompt
Defines a strict session protocol for an AI agent that implements one story per session, enforces feature branches, lock files, testing, and CI verification before marking work complete.
next_step_meta_issue
Guides a multi-step expert loop that extracts, verifies, and resolves issues until completion.
plan
Defines a read-only planning agent that investigates context and outputs actionable implementation plans without executing changes.
Base context
Defines the exact system prompt and walkthrough steps that OpenAI's GPT Builder uses to create and refine custom GPTs.
Your role
Defines a JSON-only communication protocol and step-by-step problem-solving workflow for an autonomous AI agent with tool delegation.
task-executor
Defines a strict protocol for an AI agent that reads a PRD, executes only explicitly requested tasks, and updates the PRD accordingly.
Meta-Expert
Guides a meta-expert to orchestrate multiple LLM experts for extracting PII from court documents, using a structured workflow.
Your role
Defines a JSON-only communication protocol for an autonomous AI agent that executes tasks using tools rather than discussing them.
Tools
Defines the system prompt for ChatGPT, including personality, tool usage rules, and output formatting constraints.
Instructions
Defines an eight-step cognitive reasoning cycle with adversarial self-correction, Git workflow, and coding instructions for an AI agent.
Migration Planning Agent
Guides an LLM agent to analyze a Chef or PowerShell repo and produce a structured Ansible migration plan.
How you work
Defines personality, planning, task execution, and communication conventions for a coding agent in the Codex CLI environment.
Instructions
Defines a Bash execution tool with rules for command chaining, git safety, and pull request creation, plus a JSON input schema.
Overture - Agent Plan Visualizer
Defines an interactive plan execution workflow where agents submit XML plans, get user approval, and update node statuses visually.
Chrome Browser Automation (Claude-in-Chrome)
Defines browser automation guidelines for Claude-in-Chrome MCP extension, covering GIF recording, console debugging, dialog handling, and tab lifecycle.
Identity
Defines Kiro's identity, autonomy policy, and manual workflow rules for spec-driven development without MCP tools.
claude_code_sink_instructions
Guides an AI agent to build a Quix Streams sink app that reads from Kafka and writes to a destination system.
Packerizing your config
Walks through installing Packer, authenticating with AWS, building an AMI, launching an EC2 instance, and connecting to a WordPress site.