Daily Workflow

This is the normal way to use AgentTrace during real work.

One Time Per Project

Run this once:

agenttrace init

You do not need to run init before every task.

Every Time You Start A New Task

Run:

agenttrace start "describe the task here" --tool Codex --model gpt-5

Examples:

agenttrace start "fix checkout page error" --tool Codex --model gpt-5

agenttrace start "add CSV export to reports" --tool Cursor --model claude-sonnet

After Code Changes

Run:

agenttrace snapshot

This saves the current Git evidence for the active task.

After Tests

Run your test through AgentTrace:

agenttrace add-test "npm test"

or:

agenttrace add-test "python -m unittest discover -s tests"

or, if the test was manual:

agenttrace add-test "manual QA" --no-execute --note "Checked the main user flow in the browser."

After Review

Write review notes in a file:

review-notes.md

Then run:

agenttrace add-review review-notes.md

At The End

Generate the final report:

agenttrace report

Generate or refresh the dashboard:

agenttrace dashboard

Simple Habit

For each task, try to capture at least:

1. one snapshot
2. one test result
3. one report

That is enough for a useful audit trail.