council-of-wisdom-omni

Council of Wisdom Omni

A multi-agent deliberation hub with 3 core agents and extensible extended agents.

When to Activate

Explicit (Primary — recommended)

• council: prefix — always activates
• "ask the council", "consult the council", "gather the council"

Auto-Trigger (Narrow — only for clear deliberation intent)

The council is for deliberate, multi-perspective decisions. It must NOT activate on isolated generic words such as "analyze", "compare", "think", "help me", "explain", "should I", "system", "api", "server", "right" or "concern" alone. Trigger only on intentional, decision-framed phrases:

• Decision requests: "which option should I pick", "weigh the trade-offs", "list the pros and cons", "help me decide between"
• Risk evaluation: "risk assessment", "what could go wrong", "evaluate the risks"
• Structured reasoning: "compare these options for", "trade-offs between X and Y"

When in doubt, prefer the explicit council: prefix and Do Not auto-activate.

Auto-Skip

hello, hi, hey, thanks, thank you
what time, weather, temperature
yes, no, ok, sure
define, what is
single-word "should I" / "analyze" / "compare" without a real decision framing

Architecture

Query → Explicit `council:`, or narrow auto-trigger
           │
           ▼ (if not skipped)
    ┌──────────────────┐
    │  CORE AGENTS    │ (always run)
    │ - Intent Decoder│
    │ - Risk Checker  │
    │ - Tone Designer │
    └──────────────────┘
           │
           ▼ (if extended triggered)
    ┌──────────────────┐
    │ EXTENDED AGENTS │ (included)
    └──────────────────┘
           │
           ▼ (only with user consent)
    ┌──────────────────┐
    │ WORKSPACE SKILLS │ (announced first)
    └──────────────────┘
           │
           ▼
    Enriched Response

Core Agents (Always Run)

Intent Decoder

What does the user actually want?

Risk Checker

What could go wrong?

Tone Designer

How should this feel?

Extended Agents (Included)

Calling Workspace Skills

The council can call skills/agents from your workspace when specialized knowledge is needed. This is always transparent and opt-in:

1. Before invoking any workspace skill, the council first tells you which skill it wants to call and why.
2. For sensitive or high-impact tasks, or if the delegation looks risky, it asks for your explicit consent first.
3. You can forbid delegation at any time (for example: "don't call other skills", "stay in your own scope") and the council will not invoke them.
4. The council never routes your request silently — every downstream call is announced before it happens.

Examples:

• Quran question → proposes calling quran-search-engine-mcp, gets your OK, then calls it
• GitHub question → proposes calling github-mcp, gets your OK, then calls it
• Security question → proposes the penetration-tester agent, gets your OK, then invokes it

To add workspace skills:

1. Skills go in workspace/skills/
2. Agents go in workspace/agents/
3. The council can then suggest them and, with your consent, call them when relevant

Adding Custom Extended Agents

Create a new .md file in agents/ folder:

# Your Agent Name

Trigger: only compound, decision-framed phrases

Your analysis...

Output & Transparency

By default the council shows what it is doing: which agents it consulted and the reasoning it applied. Use the council: prefix to see the full per-agent analysis. The council never delegates to another workspace skill silently — it always announces the delegation and (for sensitive/high-impact work) asks your consent first.

Simple Rules (80% of Value)

IF query contains: dangerous, high risk, worried about risking
THEN: Risk Checker flag = high

IF query contains: frustrated, angry, upset
THEN: Tone = empathetic

SEO

Keywords: multi-agent, AI router, agent council, decision support, AI deliberation, extensible hub, workspace skills

Use cases: personal AI assistant, decision making, risk assessment