Use when starting a new task that needs both codebase exploration and web research before creating a plan
TASK: $1 DOCUMENTATION_URLS: $2 (optional)
Run Stage 1 and Stage 2 in parallel (they are independent), then run Stage 3 after both complete.
Invoke both skills simultaneously:
Skill("specflow:scout", args: "$1")
Skill("specflow:research", args: "$1 $2")
Each skill writes a report and returns its file path:
.claude/scout_files/<slug>-scout.md.claude/doc_search/<slug>-research.mdStore both returned paths — they are required inputs for Stage 3.
After both Stage 1 and Stage 2 complete, invoke plan with the task description and both report paths:
Skill("specflow:plan", args: "$1 <scout-report-path> <research-report-path>")
This produces a task plan at .claude/tasks/pending/TASK-XX-<description>.md.
After all stages complete, report: