issue-pipelinelisted

# Issue Pipeline Skill You orchestrate the full lifecycle of GitHub issues through parallel sub-agents. Each phase runs all issues concurrently in isolated worktrees. ## Usage ``` /issue-pipeline <issue-numbers...> [flags] ``` **Flags:** - `--skip-plan` — Skip plan phase, go straight to implement (issues already have plans) - `--no-merge` — Stop after review, don't auto-merge There is deliberately no `--skip-review` flag: review can never be skipped, and work that fails or lacks review never merges — fixes go through a different `issue-action` sub-agent and a fresh independent review round, never through `pr-review-loop`, which is author self-check and never the Phase 3/4 judge. ## Before you start (controller session setup) The controller session runs in a normal shell, not a scheduler lane — set the environment before opening it, then run the pipeline from the controller prompt: ```powershell # 1. Build lane — quoted assignment; allowed values: worker-1 | worker-2 | verifier | verifier-2. # Never create an ad-hoc CARGO_TARGET_DIR (decision verification.cost-discipline). # Lane root: $env:FLEET_LANE_ROOT if set, else $env:LOCALAPPDATA\fleet-workstation\lanes. $laneRoot = if ($env:FLEET_LANE_ROOT) { $env:FLEET_LANE_ROOT } else { "$env:LOCALAPPDATA\fleet-workstation\lanes" } $env:CARGO_TARGET_DIR = "$laneRoot\worker-1" # 2. Machine label — appears in the taking: comment each sub-agent posts (Phase 2) $env:EDDA_SESSION_LABEL = "docs" # 3. Open the controller ses