pipeline

Pipeline — Bitbucket CI Status

All operations are wrapped in ${CLAUDE_PLUGIN_ROOT}/scripts/bb_pipeline.sh. Invoke subcommands instead of writing curl. Repo is auto-detected from git remote get-url origin.

Prerequisites

BITBUCKET_EMAIL + BITBUCKET_API_TOKEN in env. If missing, run /atlas:setup.

Subcommands

Call	Purpose
bb_pipeline.sh latest [branch]	latest pipeline for branch (default: current). Returns uuid state result commit branch.
bb_pipeline.sh get <uuid>	pipeline details
bb_pipeline.sh steps <uuid>	step table: name / state / result / duration / step_uuid
bb_pipeline.sh log <uuid> <step_uuid>	raw step log output
bb_pipeline.sh wait <uuid> [interval=75]	block until COMPLETED, print final state (use run_in_background: true)

Process

1. bb_pipeline.sh latest → grab uuid, state, commit.
2. If the returned commit does not match git rev-parse HEAD, inform the user — pipeline is for a different commit.
3. If state is PENDING or RUNNING, run bb_pipeline.sh wait <uuid> in background.
4. On COMPLETED, if result != SUCCESSFUL: bb_pipeline.sh steps <uuid> → find failed step → bb_pipeline.sh log <uuid> <step_uuid> → diagnose.

Output Handling

For log, scan the tail first (failures surface at the end). Quote exact error strings when reporting.

Escalation

For commit-range filtering, trigger-on-demand runs, or variable inspection, hit the API directly — see references/bitbucket-api.md.