feat: sequential task pipeline engine (A→B→C chain execution) #814
Description
Purpose
Chain tasks sequentially with data passing — when task A completes, automatically start task B with A's output as context. Essential for multi-step workflows like: analyze → plan → implement → test → review.
Changes
- Pipeline definition format (YAML or JSON)
- Pipeline execution engine in MCP server or skill
- Data passing between pipeline stages
- Pipeline status tracking and resume on failure
Acceptance Criteria
- Define pipelines with ordered stages
- Each stage can be an agent, skill, or shell command
- Output from stage N available as input to stage N+1
- Pipeline resumes from failed stage (not restart)
- Status tracking (which stage, progress %)