Skip to content
  1. Wiki
  2. Verification and evidence
  3. Decision log

Decision log

The decision log is a table the AI fills in as it works, one row per decision. Read it instead of the whole chat.

An analogy

Like a lab notebook: what was done, why, and what happened. The next day, the supervisor flips through it and gets it.

In pstack

/pstack:show-me-your-work keeps it. Each row has the time, stage, decision, reason, where the proof is, and the result. It's saved as a TSV file, decisions.tsv by default, and stays on your machine.

When you come back, say "/pstack:show-me-your-work walk me through what you did last night". Before it hands you the summary, a model from another family reviews it. The reply ends with a "needs attention" section. Read that section first, then the rows it points to. For a quick status check, use /pstack:ketchup. It only reads, and tells you in plain words what changed and what's waiting on you.

Common misconceptions

  • "With a log, I don't need to review": the log makes reviewing faster. You review the decisions, not a whole night of chat.

Try it: Full guide step 7: Let it run on its own

Related: Evidence (not "should be fixed"), /pstack:poteto-mode

An unofficial beginner's guide to pstack, not its official docs. pstack is made by poteto; the official repo is the source of truth for commands and behavior. The original guide source is on GitHub.

Hosted by airali. Adapted from Xiaoxiaodong's MIT-licensed guide.