What Is a Progress File (AI agents)? — AI Glossary

James Swierczewski

May 17, 2026

What it is: A progress file is an append-only text file (usually claude-progress.md or progress.txt) that captures where an AI coding agent is in a long task. Read at session start; updated at session end. The bridge between amnesiac sessions on long-running work.
Who it is for: Developers running multi-session AI coding tasks who’ve watched agents lose context between sessions.
Best if: You want a short reference on what a progress file is and what goes in one.
Skip if: You only use AI for short, single-session tasks. Want one practical AI workflow every morning? Subscribe to our free daily newsletter.

Table of Contents

What is Progress File (for AI agents)?

A progress file is a short text file at the repo root that captures where the AI coding agent is in a long task. Anthropic’s standard name is claude-progress.txt or progress.md. The file is append-only: agents read it at session start to learn what’s been tried and what worked, then add structured updates before exiting. The pattern bridges the gap between sessions on multi-day work — the agent’s working memory resets, but the filesystem doesn’t.

Why does Progress File (for AI agents) matter?

Without a progress file, every new session starts from scratch. The agent re-discovers approaches that already failed, re-litigates decisions that were already made, and loses the institutional memory of the project. Addy Osmani’s framing: “The agent itself is amnesiac, but the filesystem isn’t.” A 200-byte progress note saves hours of repeated work and prevents the worst kind of long-running-task failure: the loop where the agent keeps trying the same broken approach.

How does Progress File (for AI agents) work?

A good progress file tracks four things:

Current status — one paragraph on where the task is right now.
Tasks completed — short checklist of what’s done and verified.
Failed approaches and why — the most important section. Without this, the next session re-attempts dead ends.
Next step — the single most important thing the next session should do.

The agent reads the file as one of its first actions in any new session, alongside AGENTS.md and the feature list. The three files together form what Addy Osmani calls the “ralph-loop” pattern. See our Long-Running Claude Code Tasks guide for the full pattern.

Related terms

Learn more on Beginners in AI

Sources and further reading

Last reviewed: May 2026. AI terminology evolves quickly — verify specifics on the official source pages above.

Get Smarter About AI Every Morning

Free daily newsletter — one term, one tool, one tip. Plain English.

Free forever. Unsubscribe anytime.

Gemini Pricing: Free, Pro & Ultra

Best AI Prompts for Social Media

Do AI Detectors Work? What to Know