Turn a Patch Session Into a PR Body: A JSONL Contribution Journal

Description and use

A JSONL journal records every command and question, enabling automatic PR body generation that reflects actual outcomes and verified answers, reducing reviewer effort.

Problem

Patch contributions lack a verifiable evidence trail, leading to reviewer inefficiency.

Outcome

The journal format captures every command and question, ensuring PR bodies reflect actual outcomes and verified answers, reducing reviewer effort.

Workflow

• Create journal.py with append, run, ask, and report functions
• Use run() to execute commands and log command, exit code, duration, and last output lines
• Use ask() to log questions, sources, answers, and verification status
• After patch session, run report() to generate a PR body from the journal entries

Evidence

A patch is only as reviewable as its evidence trail.

Source

Three problems repeat in model-assisted contribution work.

Source

A run record stores a command, exit code, duration, and the last output lines.

Source

The verified flag carries the weight. It stays false until a human checks the answer against code or output.

Source