campaign

name: campaign description: Run a coordinated-leg campaign over the approved slate — the OUTER loop that partitions work into ordered, cap-bounded legs and sequences them so the global rate-limit budget is spent in bounded batches. Equivalent standalone entry to /pipeline:fullsend --campaign. Usage: /pipeline:campaign [issue_numbers...] [--manual-merge] [--spawn] [--max-bc=N] [--max-ad=N] disable-model-invocation: false allowed-tools: Read, Bash, Glob, Grep, Agent

Boot

At session start, before running any of the steps below, source the project's pipeline.config so the PIPELINE_* variables are available for the rest of this skill:

source "$(pwd)/pipeline.config" 2>/dev/null || source ./pipeline.config
# Self-resolve CLAUDE_PLUGIN_ROOT in case the env var is unset in the Bash subshell.
# Anchor via the plugin cache glob (var-independent — no chicken-and-egg dependence on
# CLAUDE_PLUGIN_ROOT to FIND the resolver). _cpr_dir is the dir prefix; literal source line.
_cpr_dir="${CLAUDE_PLUGIN_ROOT:+${CLAUDE_PLUGIN_ROOT}/}"
_cpr_dir="${_cpr_dir:-$(ls -d ${HOME}/.claude/plugins/cache/claude-pipeline-local/pipeline/*/ 2>/dev/null | sort -V | tail -1)}"
_cpr_dir="${_cpr_dir:-$(ls -d ${HOME}/.claude/plugins/cache/claude-pipeline/pipeline/*/ 2>/dev/null | sort -V | tail -1)}"
source "${_cpr_dir}scripts/_resolve-plugin-root.sh" 2>/dev/null || true

The bash code blocks referenced below resolve PIPELINE_REPO, PIPELINE_BASE_BRANCH, PIPELINE_CAMPAIGN_MAX_BC, PIPELINE_CAMPAIGN_MAX_AD, etc. from the sourced config — not from envsubst at install time.

Campaign — the coordinated-leg entry point

/pipeline:campaign runs the coordinated-leg OUTER loop over a slate of approved issues: it classifies/plans/eval-plans the ENTIRE set batched-under-caps, approves all up front, partitions the approved set into ordered per-leg batches via scripts/plan-campaign.sh (honoring PIPELINE_CAMPAIGN_MAX_BC / PIPELINE_CAMPAIGN_MAX_AD, overridable per-invocation via --max-bc=N / --max-ad=N), then runs each leg through the existing wave-by-wave execute → eval-pr → greenlight-merge machinery — advancing the base tip between legs and folding/filing campaign-surfaced bug signals at the end. This spends the global rate-limit budget in bounded batches rather than a single flat blast of the whole set.

Single source of truth — defer to fullsend's ## Campaign mode

This skill is a THIN entry point: it owns NO leg-loop prose of its own. The campaign machinery — campaign flow (a)→(e), the global-budget rule, the base-advance + usage read-out at each leg boundary, the per-leg two-layer signal dedup, the End-of-campaign fold wave (#838), End-of-campaign bug filing, the scoped halt (dependency-closure drop), and the self-mutation callout — is documented as the single source of truth in skills/fullsend/SKILL.md ## Campaign mode.

Follow that ## Campaign mode section verbatim over this skill's slate. /pipeline:fullsend --campaign enters the identical machinery — the two entry points share one copy of the leg-loop prose, so they can never drift. Carry forward all of that section's contracts by reference:

• Caps / overrides: PIPELINE_CAMPAIGN_MAX_BC (B/C-pool per-leg cap) and PIPELINE_CAMPAIGN_MAX_AD (A/D-pool per-leg cap) from the environment, overridable per-invocation via the --max-bc=N / --max-ad=N flags forwarded to scripts/plan-campaign.sh. PIPELINE_CAMPAIGN_MAX_FOLD (default 3) bounds the end-of-campaign fold wave.
• Partitioner / helpers (reused verbatim, NOT forked): scripts/plan-campaign.sh (partition, closure, aggregate-signals, fold-select) and scripts/usage-surface.sh (leg-boundary usage read-out).
• Autonomy + two-layer dedup (#863): end-of-campaign filing routes through the deterministic, NON-INTERACTIVE create-issues subset only (never the interactive brainstorming dialogue), and dedups against open issues + the running campaign-filed set both per-leg and at completion.
• Resolver-driven execute dispatch (#1056): each leg's PATH B / PATH D execute dispatch routes through scripts/resolve-execute-dispatch.sh (the single-source MODEL + split-role + carve-out resolver) exactly as fullsend's Step 6 does — inherited by reference, no logic duplicated here. Edit the resolver / fullsend Step 6, and campaign legs pick it up.

--spawn (reverts PATH C to the legacy spawn-claude.sh run-queue transport) and --manual-merge compose freely with this skill exactly as they do under /pipeline:fullsend --campaign. There is no machinery here to maintain separately — edit the canonical ## Campaign mode section in skills/fullsend/SKILL.md and both entry points pick it up.