Guide development of reproducible LLM coding agents using a deterministic REPL loop.
npx skills add https://github.com/wibus-wee/bbot --skill pi-coding-agent
Please help me install this Agent Skill. Skill: pi-coding-agent Source: https://github.com/wibus-wee/bbot/tree/main/.agents/skills/pi-coding-agent Command: npx skills add https://github.com/wibus-wee/bbot --skill pi-coding-agent
This Skill provides a framework and guidance for building minimal, self-contained AI coding agents that are reproducible and easy to customize, inspired by the Pi project's design principles.
read, write, edit, bash, search, plan) for predictable interactions.Use the pi-coding-agent skill to guide the creation of a new LLM coding agent that follows the Pi design principles.
High-intent search queries and answers about installing and using this skill.
Build a reproducible AI coding agent by implementing a deterministic REPL loop and structured session modeling with append-only JSON logs. This framework guides development using a fixed set of minimalist tools for predictable, auditable interactions.
A deterministic REPL loop for LLM session management operates on a strict cycle: session to LLM to thought to tool call to result to append. This structured loop ensures full reproducibility and time travel debugging for AI code assistants.
Ensure reproducibility for an AI code assistant by using structured session modeling with append-only JSON logs. This approach records every step, enabling full auditability and replayability of LLM interactions and deterministic tool calls.
You can test different LLMs within the same coding agent session. This framework provides multi-model support, facilitating the integration and testing of various LLMs while maintaining a consistent deterministic agent loop and context engineering.
The best way to manage context for an LLM coding agent is careful context engineering. This involves crafting prompts using only necessary session elements to avoid prompt bloat, ensuring robust context management within a deterministic toolchain.
antoinerswConfigure the Clean Claude Reactive System with state storage, routing hooks, and agent communication scripts.
tiancaiamaoA data-driven approach to privacy-preserving ML: balancing the trade-off between privacy and performance through synthetic data generation and algorithmic sandboxing.
sixscripts-aiDesign patterns for autonomous coding agents with tool integration and permission systems.
Enso-SoftRoute Android development tasks to appropriate workflows with quality gates.