do-plan-critiquelisted

# Plan Critique (War Room) ## Execution Context This skill runs inline in the context that invokes it. It is deliberately not a `context: fork` skill: the harness withholds the Agent tool from any subagent at its spawn-depth limit (three layers below the main conversation by default, and a forked skill's subagent is an ordinary subagent, not a conversation fork). An SDLC supervisor already runs each stage inside its own general-purpose Agent, so a forked critique landed exactly at that limit and could spawn no critics; the roster silently collapsed into one agent applying every lens in sequence while the report kept the war-room format. Running inline keeps the roster one layer shallower and leaves the isolation to whoever dispatched the stage. Invoke it directly from a top-level conversation and the critics are ordinary subagents; invoke it from a stage-runner Agent and they are that runner's subagents. ## Repo Context Probe If `docs/sdlc/do-plan-critique.md` exists, read it and honor its declarations; otherwise use the generic defaults described below. The context file is where a repo layers its SDLC automation onto this generic baseline: stage/status markers around the critique, the repo's mandated plan sections, force-FULL doctrine paths, a resume/roster-completion barrier (a crash-resume probe, a frozen roster manifest, a plan-hash guard, and a membership-gate CLI), a verdict-recording substrate the downstream pipeline reads, and a plan-revising lock. When the file