The micro execution loop

A roadmap step is only trustworthy when the small agent loops underneath it are observable. SoloMap uses a six-part loop to turn agent activity into project facts.

Intent

Capture what the user wanted and what this loop is supposed to change.

Judgment

Record the chosen path, affected files, tradeoffs, boundaries, and expected validation.

Action

Track what actually changed in the project, not just what the agent claimed.

Evidence

Verify with tests, rendered output, logs, diffs, screenshots, or concrete files.

Result

Classify the loop as closed, partial, failed, unverified, deviated, or follow-up required.

Attribution

Connect the evidence back to a roadmap step, user capability, product boundary, or risk.