Agent Claude Code
Architecture team — route structure, module boundaries, state management, data flow. Invoke before introducing new features or to review existing structure.
not rated 7 5mo ago A 29 tokens
original MIT
Agent Claude Code
Code-review team — coding conventions, readability, types, framework patterns. Invoke for file- or PR-level detailed review.
not rated 7 5mo ago A 28 tokens
original MIT
consultant
03Agent Claude Code
Consultant — external benchmarking, industry trends, business-model advice. Invoke when investigating competitor patterns or industry best practices.
not rated 7 5mo ago A 26 tokens
original MIT
Agent Claude Code
DB team — schema, foreign keys, access control, migrations, queries. Invoke after queries to check missing FK hints, or when adding new columns.
not rated 7 5mo ago A 35 tokens
original MIT
dev-team
05Agent Claude Code
Dev team — actual code implementation. Builds pages, API routes, components, DB migrations. The core executor invoked when the Director dispatches feature work. Each parallel instance acts independently.
not rated 7 5mo ago A 39 tokens
original MIT
director
06Agent Claude Code
Director — the persona that fronts the multi-agent harness. Acts as a virtual character whose dispatches and reports are logged to the chat room ({{ADMINHARNESSROUTE}}). Actual team calls and code work are handled by Main Claude, which reads this guide, "plays the Director," and writes to the chat room.
not rated 7 5mo ago A 67 tokens
original MIT
Agent Claude Code
Docs team — keeps CLAUDE.md / README / PRD / TASKS in sync. Invoke when checking docs after code changes, or when writing new docs.
not rated 7 5mo ago A 37 tokens
original MIT
qa-team
08Agent Claude Code
QA team — functional / scenario-based E2E and UI verification. Invoke after a feature is built, from a user-scenario perspective.
not rated 7 5mo ago A 31 tokens
original MIT
Agent Claude Code
Security team — auth, permissions, access control, signatures, XSS, CSRF, PII protection. Invoke after sensitive API / payment / auth flow changes, or before deploys.
not rated 7 5mo ago A 41 tokens
original MIT
Agent Claude Code
Verifier team — final verification after code changes. Build, regression, cross-check of other teams' reports. The mandatory gate before the Director reports "done.".
not rated 7 5mo ago A 35 tokens
original MIT
bujang-init
11Command Claude Code
Install the Harness-Bujang multi-agent system into the current project — agents, CLAUDE.md section, learning log, and extra-tool adapters.
not rated 7 5mo ago A 35 tokens
original MIT
Command Claude Code
Summarize recent chat-room activity — what each team worked on, severity counts, and open blockers.
not rated 7 5mo ago A 26 tokens
original MIT
Command Claude Code
Show the current Harness-Bujang installation status — agents, CLAUDE.md section, learning log, chat DB, and recent chat messages.
not rated 7 5mo ago A 34 tokens
original MIT
bujang-team
14Command Claude Code
Quick-dispatch shortcut — call a specific subagent without the Director's full ceremony. Useful for one-off audits.
not rated 7 5mo ago A 28 tokens
original MIT
open-chat
15Command Claude Code
Launch the Harness-Bujang chat-room viewer (KakaoTalk-style) at http://localhost:7777 so the user can watch director / team INSERTs in real time. Backgrounds the server and auto-opens the browser.
not rated 7 5mo ago A 50 tokens
original MIT
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: