AI AGENTS
BuiltThe Council & Agent Roster
A Claude Code setup with 25 role-based sub-agents and a Council command that runs five of them in parallel, then synthesises one verdict.
PROBLEM
Important decisions suffer from a single viewpoint, and one general-purpose AI session blends every role together.
WHAT I BUILT
- 25 agent definitions covering strategy, critique, operations, HR operations, web development, UI/UX, security review, SEO, automation engineering and more.
- A Council workflow that launches five agents at once (strategist, critic, creative, operator and advocate), then peer-reviews their output.
HOW IT WORKS
- The user poses a question and the Council restates it.
- Five agents analyse it in parallel, each through a different lens.
- Their outputs are reviewed against each other and merged into one recommendation with an execution plan.
TECHNOLOGY
- Claude Code
- Sub-agents
- Agent orchestration
- Prompt engineering
WHAT IT DEMONSTRATES
- Designing multi-agent workflows
- Role definition and parallel agent orchestration
- Turning AI output into an actionable plan