Chat History Summarizer

Produces a structured markdown summary a future LLM can read and act on without re-reading the full conversation. Covers session goal, actions taken, decisions made, problems and solutions, current file state, pending work, and key facts for the next session.

Output Format

Only include sections that are relevant — omit empty ones.

# Session Summary

**Date:** YYYY-MM-DD
**Branch / Repo:** <branch name and repo if known>

## Goal

One or two sentences: what was the user trying to accomplish?

## What Was Done

Ordered list of actions taken, first to last. One line per item.

- Action 1
- Action 2

## Decisions Made

Decisions that affect future work — naming, locations, approaches.

| Decision | Rationale |
| -------- | --------- |
| ...      | ...       |

## Problems and Solutions

Non-obvious issues and how they were resolved.

| Problem | Solution |
| ------- | -------- |
| ...     | ...      |

## Current State

What files were created, changed, or deleted?

- Created: `path/to/file` — purpose
- Modified: `path/to/file` — what changed
- Deleted: `path/to/file`

## Pending / Next Steps

Work discussed but not completed, or logical next actions.

- [ ] Item

## Key Facts for Next Session

Non-obvious facts a future LLM must know to avoid repeating mistakes or redundant questions.

- Fact 1

Writing Rules

Factual, not narrative. State what was done. Avoid "we explored" or "the user wanted to".

One fact per line. Bullet points are easier to parse than paragraphs.

Specific over general. File paths, command names, and exact decisions over vague descriptions.

Current State is the most important section. List every file created, modified, or deleted.

Key Facts is for gotchas. If something caused a mistake or correction, it belongs here.

Omit the obvious. Skip routine tool use, standard git operations, and anything derivable from reading the files directly.

Aim for minimum length that preserves full context. Readable in under 2 minutes by a human, parseable in one pass by an LLM.