/doctor prompt audit

Prompt Audit is a Claude Code skill, also callable through the Claude API, that examines prompts, skills, and tools for outdated or conflicting practices. It identifies anti-patterns such as instructing a model to “think step by step,” unnecessary verification rituals, and contradictory rules, then removes those patterns. The videos describe it as a way to improve model performance and reduce usage costs.

Skill Doctor is a runnable skill for Claude Code, Codex, and similar single-player agent setups. It collects past agent conversations, including transcripts, tool calls, and outputs, evaluates them in batches using rubrics such as redundant tests, verbosity, code quality, and efficiency, and produces a report card with suggested changes to skills and configuration. It is presented as part of a self-improvement workflow in which a daily agent identifies recurring failures and can propose fixes through pull requests.

cc-prune is a Python command-line tool for surgical context recovery in Claude Code sessions. It removes already-processed tool output from a Claude transcript while preserving conversation text, reasoning, thinking blocks, transcript structure, and tool-call relationships; it does not summarize or truncate the retained content. The workflow measures actual API usage, inspects transcript byte buckets, optionally audits individual results, and clears selected buckets such as tool results, inputs, attachments, or orphaned command output. Clear operations use a character-length threshold, can protect a tail of recent records, create numbered snapshots, validate invariants, and support restoration. Its validation checks include unchanged record and UUID/parent-UUID chains, byte-identical thinking signatures, and matching tool-result and tool-use identifiers. The tool operates on Claude Code JSONL transcripts under the user's .claude/projects directory and requires the session to be stopped before editing. It supports usage verification after pruning, snapshot listing and restoration, and density calculations based on measured token counts. The project states that the Claude Code transcript format is undocumented and version-sensitive, and that cleared results may need to be regenerated; it is distributed under the MIT license and requires Python 3.9 or later with no dependencies.

Andrej Karpathy Skills is a set of guidelines for Claude Code and Cursor that configures coding agents to avoid common LLM coding pitfalls. Its four principles are Think Before Coding, which requires explicit assumptions, clarification of ambiguity, and stated trade-offs; Simplicity First, which limits speculative features and unnecessary abstractions; Surgical Changes, which restricts edits to the requested scope; and Goal-Driven Execution, which turns tasks into verifiable success criteria and test-driven loops. The guidelines can be installed as a Claude Code plugin, copied into a project's CLAUDE.md file, or applied through the repository's Cursor rule. The repository is licensed under the MIT License.