← ClaudeAtlas

sprint_reviewlisted

Execute an assigned sprint unit review or the close-time conformance pass. Review the requested exact head, test high-value properties adversarially, return Major/Medium/Low findings and a recommendation to the planner, declare review-clean explicitly, or write requirement-level conformance verdicts against integrated main. Load when the boot sprint directive or a scoped task assigns review work.
jedbjorn/subfloor · ★ 15 · Code & Development · score 76
Install: claude install-skill jedbjorn/subfloor
# sprint_review Use the base `review` discipline with the sprint's exact-head and planner-routed handoff contract. ## Activate from the record Read the scoped task and board: ```sh ./sc mem message check ./sc sprint board --sprint <doc-id> ./sc mem get doc --doc <doc-id> ``` Confirm the assigned unit or conformance scope. Keep: ```text SPRINT doc=<id> reviewing=<unit or conformance> ``` Every review transition uses a durable scoped result: ```sh ./sc mem message send <planner> "$(<./sprint-result.md)" \ --kind result --sprint <doc-id> ./sc mem message sent ``` Compose message and task bodies via a file, never inline in a quoted shell string: backticks and `$()` execute before `sc` receives the body. Write the body to `./sprint-result.md`, send its content as one argument, then use `message sent` to read the stored row back and confirm its body. File findings and verdicts within the assignment. Send ruling requests to the planner with alternatives and evidence. Message IDs are scoped to their recipient. Treat an ID from another actor's inbox as provenance, never as a fetch instruction; the task row must quote the substance you need. Ask the planner for that substance when it is absent. Use IDs returned by creating writes for documents, tasks, flags, and messages. Confirm the target before an irreversible mutation. Refer to flags by `flag_id` plus a sprint-scoped label. Establish absence through a complete direct read, count, or exact-ID query. Before closing a fl