config.json reference
Path: .pipeline/config.json. Do not put secrets or tracker site URLs in this file.
workflows
Each key is a workflow name. Fields you will see:
| Field | Meaning |
|---|---|
source | text or jira |
plan_source | Artifact the planner reads (prd.md, intake.md, epic-plan.md, rca.md) |
chain | Ordered specialist tokens and @signoff:* / @waves |
classes | micro / minor / feature chains (feature-development only) |
skips | skip_pm, skip_ba, skip_architect, skip_telemetry, … |
run_tester | Force tester on (bugs) |
default_change_class | Starting class for tracker workflows |
waves.child_chain
Expanded per child listed in features/{slug}/spec-order.md. Default: developer → developer-critic.
intake.jira
enabled, key_pattern, mcp_namespaces, tools.issue / tools.search, epic_children_jql, max_children, issue_type_map, include_comments, write_back.
product
artifact_dir (default features/). readonly_agents may write only there.
Opt-in blocks
| Block | Set by |
|---|---|
test_design | knowledge init (enabled, graph_path, playwright, …) |
architecture_diagrams | plugins install archify (version v2.16.0, fallback mermaid) |
agent_observability | obs install (adapter, redact, dataset, retention) |
verify.rules
Path substring + message. Optional expect regex for obs integrity.
gates
retry_cap (default 2), require_signoff_before_deploy, require_planning_signoff_before_build, complete_after (retro-agent).
deploy
target and targets list. Must match the local-deploy script.