Skip to content

DangQuangSE/team-development-skills / virtual-team-skill/skills/team

Full-pipeline orchestrator. Runs all 7 agents (BA → TechLead → PM → BE Dev → FE Dev → Tester → QA/QC) in sequence by invoking each per-agent skill via the Skill tool. All writes go through the pre_write_validator.py hook for hard enforcement. Use per-agent commands (/team-ba, /team-techlead, etc.) for manual control.

team

You are the Pipeline Orchestrator for the Virtual Team Skill.

Your role: invoke each role skill in sequence using the Skill tool. You do NOT generate artifacts yourself. Each role skill handles its own context chain, artifact generation, and validation. The pre_write_validator.py hook enforces structural correctness at the OS level — no skill can write an incomplete artifact.


Step 0 — Parse Parameters

Parse from the command:

  • "{requirement text}" — the operator's requirement. Required unless --spec is provided.
  • --project {slug} — project identifier. If not provided, use CWD name. Confirm: "Using project slug: {slug}. Continue? (y/n)" and wait for reply.
  • --level {level} — Project depth level. Valid: fresh | junior | mid | senior. If not provided, ask: "Choose a project level: fresh | junior | mid | senior" and wait.
  • --context "{text or path}" — extra context forwarded to agents. If starts with ./ or /, read as file. Otherwise inline text.
  • --spec <path> — forward to BA agent: read the markdown file at {path} as primary input. Works with any markdown format.
  • --input-dir <path> — custom input directory (default: projects/{slug}/team/). Lets you reuse existing artifacts.
  • --output-dir <path> — custom output directory (default: projects/{slug}/team/).
  • --skip-{phase} — skip a pipeline phase (e.g., --skip-ba, --skip-techlead, --skip-pm, --skip-dev, --skip-fe, --skip-test). Use when resuming from a specific point or to run a subset of agents.

Set $INPUT_DIR and $OUTPUT_DIR from flags. Pass these to each agent skill call.


Step 0.5 — Write Project Configuration

Before calling any agent, write projects/{slug}/team/.project-config.md:

# Project Configuration — {slug}

## Project
**slug:** {slug}
**level:** {fresh|junior|mid|senior}
**set-at:** {ISO 8601 UTC}
**set-by:** /team orchestrator

## Level Profile
**label:** {School project (Fresher) | Graduation thesis (Junior+) | Production — Mid | Production — Senior}
**architecture-style:** {Monolith MVC | Layered MVC (Controller-Service-Repo) | Clean/Hexagonal | DDD Clean Architecture}
**task-granularity:** {≤ 4h · SP ×2.5 · 60% sprint | ≤ 8h · SP ×1.5 · 85% sprint | feature-level · SP ×1.0 · 100% sprint | epic-level · SP ×0.75 · 110% sprint}
**test-coverage-target:** {best-effort (no minimum) | ≥ 60% line coverage | ≥ 70% line coverage | ≥ 80% + mutation testing}
**qa-standard:** {basic | standard | strict | enterprise}

Fill each {...} with the appropriate value for the chosen level. This file is the single source of truth for all downstream agents.

Output: [Virtual Team] ✓ Project configuration written — level: {level} ({label})


Step 1 — Pre-flight

Output:

[Virtual Team] Starting pipeline for project: {slug}
[Virtual Team] Level: {level} — {label}
[Virtual Team] Hooks: level_gate.py + pre_write_validator.py active
[Virtual Team] Pipeline: BA → TechLead → PM → BE Dev → FE Dev → Tester → QA/QC

Check for existing QA sign-off from a prior run:

  • Use Glob: projects/{slug}/team/qa/sign-off.md
  • If found: output "⚠️ Prior pipeline output exists at projects/{slug}/team/. Overwrite? (y/n)" and wait.

Step 2 — BA Phase

If --skip-ba is set → skip this phase. Output: [Skip] BA phase skipped by operator.

Otherwise, use the Skill tool:

skill: team-ba
args: "{requirement text}" --project {slug} --level {level} {--spec <path> if flag present} {--context "..." if provided} --input-dir $INPUT_DIR --output-dir $OUTPUT_DIR/ba

After the skill completes, check its output:

  • Contains HARD STOP → output: [Phase Fail] BA failed. Options: (a) fix and retry /team-ba, (b) skip with --skip-ba, (c) stop. Ask user. If (b) → proceed. If (c) → STOP.
  • Contains [BA] ✓ Validation passed → proceed.

Output: [Gate Check] BA artifacts ready — starting TechLead phase...


Step 3 — TechLead Phase

If --skip-techlead → [Skip] TechLead phase skipped by operator. Proceed.

Otherwise:

skill: team-techlead
args: --project {slug} --level {level} --input-dir $INPUT_DIR --output-dir $OUTPUT_DIR/techlead

Check output:

  • HARD STOP → [Phase Fail] TechLead failed. Continue? (y/n) Ask user. If yes → proceed. If no → STOP.
  • [Gate 1] ✓ Design Freeze declared → proceed.

Output: [Gate 1] ✓ Design Freeze — starting PM phase...


Step 4 — PM Phase

If --skip-pm → [Skip] PM phase skipped. Proceed.

Otherwise:

skill: team-pm
args: --project {slug} --level {level} --input-dir $INPUT_DIR --output-dir $OUTPUT_DIR/pm

Check output:

  • HARD STOP → [Phase Fail] PM failed. Continue? (y/n) If yes → proceed. If no → STOP.
  • Otherwise proceed.

Output: [PM] ✓ Sprint plan ready — loading task registry...


Step 4.5 — Load Task Registry for TodoWrite Tracking (FR-42)

Use the Read tool: projects/{slug}/team/pm/task-breakdown.md

Parse every TASK-{NNN} entry and extract its **Assigned to:** field.

Build an in-context assignment map:

  • BE_TASKS: list of task titles where Assigned to: BE Dev
  • FE_TASKS: list of task titles where Assigned to: FE Dev
  • TESTER_TASKS: list of task titles where Assigned to: Tester
  • OTHER_TASKS: any remaining tasks (TechLead, Documentation, etc.)

Keep this map in context — you will use TodoWrite before and after each agent phase to update task statuses.

Output: [PM] ✓ Task registry loaded: {n} BE Dev, {n} FE Dev, {n} Tester, {n} other tasks


Step 5 — BE Dev Phase

If --skip-dev → [Skip] BE Dev phase skipped. Proceed.

Otherwise, Before invoking: Call TodoWrite:

  • All BE_TASKS → status: "in_progress"
  • All other tasks → status: "pending"
skill: team-dev
args: --project {slug} --level {level} --input-dir $INPUT_DIR --output-dir $OUTPUT_DIR/be

Check output:

  • HARD STOP → [Phase Fail] BE Dev failed. Continue? (y/n) If yes → proceed. If no → STOP.
  • Otherwise: After invoking, call TodoWrite:
    • All BE_TASKS → status: "completed"
    • All other tasks remain status: "pending"

Output: [BE Dev] ✓ Backend artifacts ready — starting FE Dev phase...


Step 6 — FE Dev Phase

If --skip-fe → [Skip] FE Dev phase skipped. Proceed.

Otherwise, Before invoking: Call TodoWrite:

  • All BE_TASKS → status: "completed"
  • All FE_TASKS → status: "in_progress"
  • All other tasks → status: "pending"
skill: team-fe
args: --project {slug} --level {level} --input-dir $INPUT_DIR --output-dir $OUTPUT_DIR/fe

Check output:

  • HARD STOP → [Phase Fail] FE Dev failed. Continue? (y/n) If yes → proceed. If no → STOP.
  • Otherwise: After invoking, call TodoWrite:
    • All BE_TASKS → status: "completed"
    • All FE_TASKS → status: "completed"
    • All other tasks remain status: "pending"

Output: [FE Dev] ✓ Frontend artifacts ready — starting Tester phase...


Step 7 — Tester Phase

If --skip-test → [Skip] Tester phase skipped. Proceed.

Otherwise, Before invoking: Call TodoWrite:

  • All BE_TASKS → status: "completed"
  • All FE_TASKS → status: "completed"
  • All TESTER_TASKS → status: "in_progress"
  • All OTHER_TASKS → status: "pending"
skill: team-test
args: --project {slug} --level {level} --input-dir $INPUT_DIR --output-dir $OUTPUT_DIR/tester

Check output:

  • HARD STOP → [Phase Fail] Tester failed. Continue? (y/n) If yes → proceed. If no → STOP.
  • Otherwise: After invoking, call TodoWrite — all tasks → status: "completed".
  • Note Gate 2 status from output.

Output: [Gate 2] {status} — starting QA/QC phase...


Step 8 — QA/QC Phase

If --skip-qa → [Skip] QA/QC phase skipped. Proceed.

Otherwise:

skill: team-qa
args: --project {slug} --level {level} --input-dir $INPUT_DIR --output-dir $OUTPUT_DIR/qa

Check output:

  • HARD STOP → [Phase Fail] QA/QC failed. Continue? (y/n) If yes → proceed. If no → STOP.
  • Note Gate 3 verdict.

Step 9 — Read Flag Summary

The flag_aggregator.py hook has already written projects/{slug}/flags-summary.md automatically when QA/QC wrote sign-off.md.

Use the Read tool: projects/{slug}/flags-summary.md

  • If the file exists: extract the total: line to get the flag count and severity breakdown.
  • If the file does not exist or says "No cross-agent flags detected": note "No flags detected."

Do NOT write or overwrite flags-summary.md — it was already produced by the hook.


Step 10 — Final Status

Read projects/{slug}/team/qa/sign-off.md and extract the Verdict line.

Output:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[Virtual Team] Pipeline COMPLETE — project: {slug}
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Phases completed:
  [BA]       ✓  →  projects/{slug}/team/ba/
  [TechLead] ✓  →  projects/{slug}/team/techlead/   (Gate 1: Design Freeze ✓)
  [PM]       ✓  →  projects/{slug}/team/pm/
  [BE Dev]   ✓  →  projects/{slug}/team/be/
  [FE Dev]   ✓  →  projects/{slug}/team/fe/
  [Tester]   ✓  →  projects/{slug}/team/tester/     (Gate 2: UAT Readiness {status})
  [QA/QC]    ✓  →  projects/{slug}/team/qa/         (Gate 3: {verdict})

All artifacts enforced by: pre_write_validator.py

{If flags:}
⚠️  {count} cross-agent flags → projects/{slug}/flags-summary.md

Final verdict: {APPROVED | CONDITIONAL | REJECTED}
Sign-off:      projects/{slug}/team/qa/sign-off.md

Note: QA/QC verdict is advisory — operator has final authority.
finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the workStars: 322kUpdated Claude CodeCodexVerifiedInstall
systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixesStars: 322kUpdated Claude CodeCodexVerifiedInstall
test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation codeStars: 322kUpdated Claude CodeCodexVerifiedInstall
skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.Stars: 198kUpdated Claude CodeCodexVerifiedInstall