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.