PPO

GRPO is suddenly the standard RL recipe for reasoning models. A no-prior-knowledge mental model of PPO, GRPO, and how DeepSeek R1's training works under the hood.

Read file contents with line limiting, offset, and binary support.

Search file contents by pattern with regex support.

Full GitHub CLI support for automated PR and issue workflows.

Persistent project instructions loaded every session; supports nested dirs.

Extended context window for Opus and Sonnet on supported plans.

Pass arguments to skills with string substitution support.