blindspotlisted

# blindspot Analyze gaps between the user's framing and repository evidence. The output is only a grounded Blindspot Report. ## Evidence Capture the user's goal, unfamiliar area, and stated beliefs. Then dispatch agent `scout` in `blindspot` mode with those inputs. Ask for how the area works, relevant standards and constraints, coupling or prior attempts, and git signals. Treat the report as evidence, not as an implementation recommendation. ## Analysis Compare the framing against the evidence. Include only findings grounded in evidence or directly implied by the framing; omit empty categories. | Category | Include | | --- | --- | | Unknown unknowns | Relevant concerns the framing did not account for | | False assumptions | A stated belief contradicted by evidence | | Hidden constraints | Coupling, standards, compatibility, or prior-art limits | | Missing questions | Questions whose answers materially change the direction | ## Output ```markdown ## Blindspot Report Goal: {the user's goal} Area: {unfamiliar area} ### Unknown Unknowns - {finding — evidence} ### False Assumptions - Believed: {belief} Actual: {evidence} ### Hidden Constraints - {constraint — evidence} ### Missing Questions 1. {question that materially changes the direction} ### Evidence Limits - {unsearched area or uncertainty} ``` Keep the analysis read-only. Do not propose a design, plan, implementation, brainstorm prompt, work-item artifact, or next-workflow handoff.