LIFEAI/rdc:workitems

⚠️ OUTPUT CONTRACT (READ FIRST): guides/output-contract.md Checklist-only output. No tool-call narration. No raw MCP/JSON/log dumps. One checklist upfront, updated in place, shown again at end with a 1-line verdict.

If dispatching subagents or running as a subagent: read {PROJECT_ROOT}/.rdc/guides/agent-bootstrap.md first (fallback: {PROJECT_ROOT}/.rdc/guides/agent-bootstrap.md).

Sandbox contract: This skill honors RDC_TEST=1 per guides/agent-bootstrap.md § RDC_TEST Sandbox Contract. Destructive external calls short-circuit under the flag. Supabase work item writes (insert_work_item, update_work_item_status, bump_epic_version) are skipped under RDC_TEST=1.

rdc:workitems — Work Item Management

When to Use

• Project lead says "create a ticket", "add to backlog", "log this bug", "update status"
• An agent needs to create or update epics/tasks during a build session
• Checking what work is open before starting a new epic

Arguments

• rdc:workitems epics — list open epics
• rdc:workitems add <title> --epic <id> — create a task under an epic
• rdc:workitems update <id> --status <status> — update a work item's status
• rdc:workitems done <id> — mark a task complete
• rdc:workitems list <epic-id> — list tasks in an epic

Rules

1. Epic first, always. Never create a task without a parent epic.
2. Label everything. Minimum one system label per item.
3. Check first. get_open_epics() before creating anything new.
4. Link design docs. Put .rdc/plans/ or .rdc/research/ paths in descriptions (fallback: .rdc/plans/ / .rdc/research/).

Read Epics

SELECT get_open_epics();
SELECT get_open_epics('urgent');
SELECT get_open_epics(p_label_filter := 'custom-label');

Create Epic

SELECT insert_work_item(
  p_title       := 'EPIC: Clear descriptive title',
  p_description := 'What and why. Reference .rdc/plans/<n>.md if applicable.',
  p_item_type   := 'epic',
  p_priority    := 'high',
  p_labels      := ARRAY['system-label'],
  p_source      := 'planning'
);

Create Task

SELECT insert_work_item(
  p_title       := 'Specific actionable task title',
  p_description := 'What: <deliverable>
Where: <files to create/modify>
Agent type: frontend | backend | data | design | infra | content | cs2
Guide: .rdc/guides/<type>.md (fallback: .rdc/guides/<type>.md)
Design doc: .rdc/plans/<n>.md (fallback: .rdc/plans/<n>.md, if exists)
Depends on: <other task title if sequential>
Est: <hours>',
  p_parent_id   := '<epic-uuid>'::uuid,
  p_item_type   := 'task',
  p_priority    := 'high',
  p_labels      := ARRAY['system-label'],
  p_estimated_hours := 2,
  p_source      := 'planning'
);

Update Status

SELECT update_work_item_status('<uuid>'::uuid, 'in_progress');
SELECT update_work_item_status('<uuid>'::uuid, 'review', '["Implementation complete; ready for validator"]'::jsonb, '<agent-session-id>', 'agent');
SELECT update_work_item_status('<uuid>'::uuid, 'done', '["Validator verified implementation report, CodeFlow post, and checklist evidence"]'::jsonb, '<validator-session-id>', 'validator');
SELECT update_work_item_status('<uuid>'::uuid, 'blocked', '["Why it is blocked"]'::jsonb);

Non-epic done is a validator-only close and requires an existing implementation_report.codeflow_post, checked required checklist evidence, and originating-agent tick audit.

Read Tasks in an Epic

SELECT get_work_items_by_epic('<epic-uuid>'::uuid);
SELECT get_work_items_by_epic('<epic-uuid>'::uuid, 'todo');

Bump Epic Version

SELECT bump_epic_version('<epic-uuid>'::uuid, '0.2.0', 'What changed', 'planning');

Valid Values

| Field | Values | |-------|--------| | status | todo in_progress blocked review done archived | | priority | urgent high normal low | | item_type | epic task subtask bug spike |

System Labels

| Label | When | |-------|------| | project-a | Your-app-specific work | | project-b | Another app work | | cs2 | Core paradigm packages | | hail | Grammar, DSL compiler | | pal | Moment windows, memory | | virtue | Virtue engine, coherence | | marketing | Outreach, campaigns | | media | Assets, R2, image pipeline | | infrastructure | CI/CD, deployment | | ui | Component library | | data | Schema, migrations | | content | Copy, messaging | | website | Public-facing sites |

Agent Type → Guide File Reference

| Type | Guide | Use For | |------|-------|---------| | frontend | .rdc/guides/frontend.md | React, pages, UI, Tailwind | | backend | .rdc/guides/backend.md | API routes, database, auth | | data | .rdc/guides/data.md | Migrations, schema, RPC | | design | .rdc/guides/design.md | Brand, palette, OG images | | infra | .rdc/guides/infrastructure.md | CI/CD, deploy, DNS | | content | .rdc/guides/content.md | Copy, messaging, tone | | cs2 | .rdc/guides/cs2.md | CS 2.0 paradigm | | hail | .rdc/guides/cs2.md + packages/hail/CLAUDE.md | Grammar, DSL | | viz | .rdc/guides/frontend.md + design.md | Custom visualizations |

What NOT to Do

• Never raw INSERT/UPDATE against work items — RPC functions only
• Never create tasks without a parent epic
• Never leave labels empty
• Never write vague titles ("Fix stuff") — be specific
• Never put design intention in the task — put it in .rdc/research/ (fallback: .rdc/research/) and link