Workflows
AgentHub is opinionated about the order you run things. The recommended flow trades a few minutes of thinking upfront for hours of wasted agent spend later.
The standard flow
1. /hub:brainstorm <idea> — explore options (MEDIUM)
2. /hub:plan <feature> — turn a direction into a plan file (MEDIUM)
3. /hub:create <app> — scaffold a greenfield app (HEAVY)
or /hub:enhance <change> — add/update features in an existing app (HEAVY)
4. /hub:test generate <module> — generate tests for the new code (MEDIUM)
or /hub:test — run existing tests (LIGHT)
5. /hub:deploy staging — pre-flight + deploy (HEAVY)
then /hub:deploy productionThe same flow in Codex: swap each /hub: for @hub and drop the colon. Every step behaves the same way on both platforms.
Why this order?
- Brainstorm first. No code. Just options. This is the cheapest step and often kills half your original idea — good.
- Plan before building. Writing
docs/PLAN-<slug>.mdtakes ~20k tokens, but the HEAVYcreatethat follows can spend 10× more — with a plan, that spend is directed. - Create or enhance, never both.
createis for greenfield.enhanceloads project context and scopes the smallest diff that satisfies the change. - Tests gate deploys.
test generateis MEDIUM because it writes code;testwithout args just runs what exists (LIGHT). - Deploy twice. Staging first, production second. Rollback is a subcommand:
/hub:deploy rollback.
Shortcut flows
Bug fix sprint
/hub:debug <symptom> → /hub:test → doneNo brainstorm, no plan. debug runs systematic-debugging, proposes a fix, applies it; test confirms nothing regressed.
UI audit and redesign
/hub:ui-ux-pro-max <target page or component>HEAVY. Dispatches hub:frontend-specialist plus 3 design skills (web-design-guidelines, frontend-design, tailwind-patterns).
Multi-domain heavy lift
/hub:orchestrate migrate auth from sessions to JWT across API + frontend + mobileTwo-phase pipeline. Phase 1: hub:orchestrator writes a plan spanning ≥3 agents and asks for approval. Phase 2: the approved plan executes with clear hand-offs.
Related
- Tiers and approval gates — the mechanics of LIGHT / MEDIUM / HEAVY.
- Commands reference — every command and its full signature.
- Usage ledger — how to audit where your tokens went.