From jaan-to
Assembles role-based AI agent teams to ship ideas from concept to production using tracks like fast, full, TDD, sprint, and custom roles. Use for multi-role orchestration and delivery.
npx claudepluginhub parhumm/jaan-to --plugin jaan-toThis skill is limited to using the following tools:
> Assemble a virtual company of AI teammates — each a role — to ship an idea from concept to production.
Creates isolated Git worktrees for feature branches with prioritized directory selection, gitignore safety checks, auto project setup for Node/Python/Rust/Go, and baseline verification.
Executes implementation plans in current session by dispatching fresh subagents per independent task, with two-stage reviews: spec compliance then code quality.
Dispatches parallel agents to independently tackle 2+ tasks like separate test failures or subsystems without shared state or dependencies.
Assemble a virtual company of AI teammates — each a role — to ship an idea from concept to production.
${CLAUDE_PLUGIN_ROOT}/skills/team-ship/roles.md - Role definitions (CRITICAL — teammate roster)${CLAUDE_PLUGIN_ROOT}/docs/extending/team-ship-reference.md - Spawn prompts, dependency algo, schemas$JAAN_CONTEXT_DIR/tech.md - Tech stack context$JAAN_CONTEXT_DIR/config.md - Project configuration$JAAN_TEMPLATES_DIR/jaan-to-team-ship.template.md - Orchestration log template$JAAN_LEARN_DIR/jaan-to-team-ship.learn.md - Past lessons (loaded in Pre-Execution)${CLAUDE_PLUGIN_ROOT}/docs/extending/language-protocol.md - Language resolution protocolArguments: $ARGUMENTS
| Argument | Effect |
|---|---|
[initiative] | Idea to build (required unless --detect or --resume) |
--track fast | 8-skill fast track: PM → Backend → Frontend → QA → DevOps |
--track full | 20-skill full track: all roles, all design steps (default) |
--track tdd | TDD track: qa-test-cases → qa-tdd-orchestrate → qa-test-mutate → qa-quality-gate |
--track sprint | Sprint track: reads sprint plan artifact from pm-sprint-plan, executes dynamic skill queue |
--detect | Detect audit mode: 5 parallel auditors → detect-pack |
--roles role1,role2 | Custom role selection from roles.md |
--dry-run | Display planned team structure without spawning |
--resume | Resume from last checkpoint |
MANDATORY — Read and execute ALL steps in: ${CLAUDE_PLUGIN_ROOT}/docs/extending/pre-execution-protocol.md
Skill name: team-ship
Execute: Step 0 (Init Guard) → A (Load Lessons) → B (Resolve Template) → C (Offer Template Seeding)
Also read context files if available:
$JAAN_CONTEXT_DIR/tech.md — Tech stack for teammate context$JAAN_CONTEXT_DIR/config.md — Project configurationRead and apply language protocol: ${CLAUDE_PLUGIN_ROOT}/docs/extending/language-protocol.md
Override field for this skill: language_team-ship
Language exception: Generated code, skill commands, YAML, and role names remain in English.
ultrathink
Use extended reasoning for:
Agent Teams enabled? Read jaan-to/config/settings.yaml for agent_teams_enabled.
If false or missing:
"Agent Teams is not enabled. To use team-ship:
- Add
agent_teams_enabled: truetojaan-to/config/settings.yaml- Set environment variable:
CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1- Restart Claude Code session" STOP — do not proceed.
Resume mode? If --resume flag:
$JAAN_OUTPUTS_DIR/team/ for most recent checkpoint.yamlinitiative text (everything not a flag)--track fast, --track full (default), --track sprint, --detect, or --roles--dry-run flagIf --track sprint:
SPRINT_PLAN=$(ls -t $JAAN_OUTPUTS_DIR/pm/sprint-plan/*/*.md 2>/dev/null | head -1)
/pm-sprint-plan first."queue array and closing_skillsReference: See
${CLAUDE_PLUGIN_ROOT}/docs/extending/team-ship-reference.mdsection "Sprint Track Execution" for plan consumption and verification gate logic.
${CLAUDE_PLUGIN_ROOT}/skills/team-ship/roles.md--roles specified: filter to only those roles--detect: select only detect-* rolesFor each selected role from roles.md:
Reference: See
${CLAUDE_PLUGIN_ROOT}/docs/extending/team-ship-reference.mdsection "Dependency Graph Resolution Algorithm" for phase grouping logic.
Read $JAAN_CONTEXT_DIR/tech.md (if exists) — summarize in 2-3 lines for spawn prompts.
This summary is injected into each teammate's spawn prompt as {tech_context_summary}.
If --dry-run flag is set:
Reference: See
${CLAUDE_PLUGIN_ROOT}/docs/extending/team-ship-reference.mdsection "Dry-Run Display Format" for the output template.
Display team plan: roster, phases, dependency graph, token estimate.
Write plan to $JAAN_OUTPUTS_DIR/team/{id}-{slug}/plan.md.
STOP — do not spawn teammates.
Present to user:
TEAM COMPOSITION
────────────────
Track: {track} ({skill_count} skills)
Initiative: "{initiative}"
Teammates ({count}):
{role}: {skills_list} [{model}]
...
Phases:
Phase 1: {roles} → PRD approval gate
Phase 2: {roles} (parallel)
Phase 3: {roles} (parallel)
Phase 4: Verify + Changelog (lead)
"Assemble this team and begin? [y/n/edit]"
If edit: let user adjust roles, track, or model choices. Rebuild roster.
Do NOT proceed without explicit approval.
If --detect mode: skip to Phase 2 (Detect Mode).
Reference: See
${CLAUDE_PLUGIN_ROOT}/docs/extending/team-ship-reference.mdsection "PM Teammate Prompt" for the full prompt template.
{initiative}, {slug}, {tech_context_summary}When PM completes:
"The PM has drafted the PRD. Review it at: {prd_path} Approve to spawn the build team? [y/n/edit]"
If no: provide feedback to PM teammate, PM revises, repeat. If yes: shut down PM teammate (free context), proceed to Phase 2.
Update checkpoint: phase=1, pm=done, artifacts.prd_path, artifacts.entities.
Determine Phase 2 roles from roster (those with Phase=2).
For each Phase 2 role:
Reference: See
${CLAUDE_PLUGIN_ROOT}/docs/extending/team-ship-reference.mdsection "{Role} Teammate Prompt" for each role's template.
{prd_path}, {stories_path}, {entities}, {slug}For --detect mode, instead spawn all detect-* roles from roster in parallel.
While Phase 2 teammates are active:
When all Phase 2 roles report done:
--detect mode: lead runs /jaan-to:detect-pack to consolidate. Skip to Phase 4 wrap-up.Update checkpoint: phase=2, role statuses.
If --track tdd selected, execute this specialized pipeline instead of standard Phase 2-3:
If qa-quality-gate skill is not available: skip with explicit warning: "Quality gate skill not available -- skipping composite scoring. Run qa-test-run coverage report as fallback."
tdd-writer role:
tdd-implementer role:
Update checkpoint: track=tdd, pipeline stages.
Lead executes directly (not a teammate — integration touches multiple output dirs):
/jaan-to:dev-project-assemble — wire scaffolds together/jaan-to:dev-output-integrate — copy to project locationsIf integration fails: update checkpoint (status=paused), present error to user.
After integration succeeds, spawn Phase 3 roles from roster:
qa-test-generate → qa-test-runqa-test-generate → qa-test-run → qa-contract-validate "{api_contract_path}"Reference: See
${CLAUDE_PLUGIN_ROOT}/docs/extending/team-ship-reference.mdsection "DevOps Teammate Prompt" and "Security Teammate Prompt".
Wait for all Phase 3 teammates to complete. Monitor same as Step 9.
Update checkpoint: phase=3, role statuses.
Lead runs directly:
/jaan-to:dev-verify — build + runtime validation/jaan-to:release-iterate-changelog — generate changelogWrite to $JAAN_OUTPUTS_DIR/team/{id}-{slug}/log.md using template.
Include: initiative, track, team roster, phase timeline, skill outputs, test results, final status.
Update checkpoint: phase=4, status=completed.
"Team work complete. Orchestration log: {log_path} All outputs in: $JAAN_OUTPUTS_DIR/"
"Any feedback on the team orchestration? [y/n]"
If yes: run /jaan-to:learn-add team-ship "{feedback}"