Claude system prompts as a git timeline

2026-04-18 · Source: Simon Willison's Weblog · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, quick

Summary

Anthropic's historical system prompts for its Claude models, originally published as a monolithic Markdown file, have been converted into a structured, git-based repository. This new format organizes prompts by model, family, and revision, assigning timestamped commits to each change. This allows researchers and developers to use standard git commands like `git log`, `diff`, and `blame` to track the evolution of prompts, compare specific versions, and identify when changes were introduced. This transformation eliminates the need for manual parsing of the original Markdown, providing a more efficient and granular way to analyze prompt history.

Key takeaway

For AI engineers and researchers analyzing large language model behavior, understanding prompt evolution is critical. You should consider adopting a git-based versioning system for your own system prompts, even for internal documentation, to easily track changes, compare versions, and attribute modifications over time, enhancing reproducibility and debugging.

Key insights

Git provides a powerful framework for tracking and analyzing the evolution of system prompts.

Principles

• Granular versioning aids analysis.
• Structured data enables tooling.

Method

Convert monolithic prompt documentation into separate, timestamped files per model/family, then commit to a git repository to enable version control features.

In practice

• Use `git diff` to compare prompt versions.
• Use `git blame` to attribute prompt changes.

Topics

• Claude System Prompts
• Git Timeline
• Prompt Engineering
• Anthropic
• Version Control

Code references

• simonw/research

Best for: AI Engineer, NLP Engineer, Research Scientist, AI Scientist, Prompt Engineer, Machine Learning Engineer

Related on AIssential

• v2.1.61 — Claude Code continuously refines its platform with new features, performance boosts, and critical bug fixes across numerous versions.
• AI Prompt Engineering Guide: ChatGPT, Claude & Others — Effective prompt engineering transforms AI from a basic chatbot into a high-performance strategic assistant by providing structured direction.
• Changes in the system prompt between Claude Opus 4.6 and 4.7 — Anthropic's Claude Opus 4.7 system prompt updates enhance safety, user experience, and tool integration.
• LAI #123: Claude Code’s Codebase Was Accidentally Leaked — Effective LLM prompting and agent design require careful consideration of example order, tool context, and subagent delegation.
• Lessons from building Claude Code: Prompt caching is everything — Prompt caching, based on prefix matching, is essential for cost-effective, low-latency agentic AI applications.
• Top 5 GitHub Repositories to get Free Claude Code Skills (1000+ Skills) — Leveraging shared Claude Code skills from GitHub significantly enhances AI assistant capabilities and workflow automation.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Simon Willison's Weblog.