Skip to main content

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:

FieldMeaning
sourcetext or jira
plan_sourceArtifact the planner reads (prd.md, intake.md, epic-plan.md, rca.md)
chainOrdered specialist tokens and @signoff:* / @waves
classesmicro / minor / feature chains (feature-development only)
skipsskip_pm, skip_ba, skip_architect, skip_telemetry, …
run_testerForce tester on (bugs)
default_change_classStarting 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

BlockSet by
test_designknowledge init (enabled, graph_path, playwright, …)
architecture_diagramsplugins install archify (version v2.16.0, fallback mermaid)
agent_observabilityobs 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.