Dated additions, corrections, and the next step for your tools.
/ Recorded run
A coding agent read the 89-word subagents brief, built a read-only code reviewer, and the reviewer reported nine findings it could not edit. The page shows the session log, the files, the time and the cost; the brief is now free.
Download the sample project and the two prompts, run them in your own Claude Code, and compare your log with ours.
Watch it run/ Addition
Submitted videos now have a private status, a completed Markdown download, and a way to tell us whether the task worked.
Open your member page after checkout. Earlier email-only requests need to be linked by support.
Open your member page/ Worked example
The sourced-summary example uses our actual plan and service notes and includes the saved answer and reusable project memory.
Compare the output with the two supplied notes, then use the same instructions with your own material.
Try the complete example/ Correction
The working-check starter states that its Python checker validates handoff structure and declared status. A passing record alone cannot prove the task ran.
Run the negative fixtures and inspect the evidence. Verify hook execution in your own host before relying on it.
Read the check and host instructions/ Addition
The free project-memory starter is available alongside the sourced-summary and working-check starters.
Save confirmed decisions and unresolved questions, then use the evidence-wiki pack when you need a larger workflow.
Open project memory