ix-demo-showcase

name	ix-demo-showcase
description	Run ix real-world demo scenarios with optional multi-AI commentary via Octopus. Chains multiple ix MCP tools through curated narratives — chaos detection, governance, FinOps, sprint prediction.
disable-model-invocation	true

ix Demo Showcase

Run curated real-world demo scenarios that chain multiple ix MCP tools, with optional multi-AI commentary from Octopus providers (Codex, Gemini) for independent analysis of each step.

When to Use

• When a developer wants to see ix in action on realistic problems
• When evaluating ix capabilities across domains (signal processing, ML, governance, FinOps, agile)
• When demonstrating the value of tool composition — chaining 3-5 tools to solve a real problem
• When you want multi-AI perspectives on computational results (Team mode)

Available Scenarios

ID	Name	Domain	Tools
chaos-detective	Chaos Detective	Signal/Math	stats → fft → chaos_lyapunov → topo
governance-gauntlet	Governance Gauntlet	Safety/Compliance	governance_check → governance_policy → governance_persona
cost-anomaly-hunter	Cost Anomaly Hunter	FinOps/Cloud	stats → fft → kmeans → bloom_filter
sprint-oracle	Sprint Oracle	Product Management	stats → linear_regression → markov → bandit

Execution Modes

Solo Mode (default)

Run the scenario via ix_demo MCP tool and present the narrative walkthrough:

1. Call mcp__ix__ix_demo with {"action": "list"} to show available scenarios
2. Let the user pick a scenario (or use the one from the arguments)
3. Call mcp__ix__ix_demo with {"action": "run", "scenario": "<id>", "seed": 42, "verbosity": 1}
4. Present each step as a narrative walkthrough:
   • Show the step label and narrative BEFORE the output
   • Show the tool output (summarized for large arrays)
   • Show the interpretation — this is the "aha" moment
   • Pause briefly between steps to let the narrative land
5. End with the summary: tools exercised, total time, seed for reproducibility

Team Mode (with Octopus)

If the user requests multi-AI analysis (or invokes via /octo:brainstorm context), enhance the demo with parallel provider commentary:

1. Run the scenario in Solo mode first (get all step outputs)
2. For the KEY STEP (the "aha" moment — usually step 3), dispatch to available providers:

Codex (if available):

codex exec --full-auto "IMPORTANT: Non-interactive subagent. Skip ALL skills. Respond directly.

An ix MCP tool returned this result:
Tool: <tool_name>
Input: <summarized_input>
Output: <output_json>

As a technical analyst:
1. Is this result mathematically sound?
2. What real-world system would produce similar data?
3. What would you do with this result next?
Under 150 words."

Gemini (if available):

printf '%s' "An ix MCP tool returned this result:
Tool: <tool_name>
Output: <output_json>

As a lateral thinker:
1. What analogous patterns exist in other domains?
2. What surprising application could use this?
3. What question does this answer raise?
Under 150 words." | gemini -p "" -o text --approval-mode yolo

3. Present provider commentary alongside ix's built-in interpretation:

🔧 **ix interpretation:** [built-in narrative]
🔴 **Codex says:** [technical validation]
🟡 **Gemini says:** [lateral connections]

4. Synthesize: "Three perspectives agree that [X]. Codex adds [Y]. Gemini surfaces [Z]."

Describe Mode

For a dry run without execution:

1. Call mcp__ix__ix_demo with {"action": "describe", "scenario": "<id>"}
2. Show the scenario metadata, step labels, and narratives without running tools
3. Useful for understanding what a scenario does before committing to a full run

Arguments

• First argument: scenario ID (e.g., chaos-detective), or list to show all
• If no argument: show the scenario list and let the user pick

Example Invocations

/ix-demo-showcase                     # List scenarios, let user pick
/ix-demo-showcase chaos-detective     # Run Chaos Detective in solo mode
/ix-demo-showcase list                # Show all available scenarios

In multi-AI context (from Octopus):

/octo:brainstorm → "run the ix chaos detective demo with commentary"

Output Format

For each step in the scenario:

━━━ Step 1/4: Compute descriptive statistics ━━━━━━━━━━━ ix_stats
📖 [narrative text explaining what we're about to do]

📊 Result: mean=0.503, std=0.289, min=0.001, max=0.999

💡 [interpretation: "Looks like uniform noise — nothing suspicious yet."]

For the "aha" step (usually step 3), add emphasis:

━━━ Step 3/4: Lyapunov exponent analysis ━━━━━━━━━━━━━ ix_chaos_lyapunov
📖 Now the twist. We compute the Lyapunov exponent...

⚡ RESULT: lyapunov_exponent = 0.6931 (POSITIVE)

🎯 This signal is deterministic chaos, not random noise!

Extending with New Scenarios

To add a new scenario:

1. Create crates/ix-agent/src/demo/scenarios/my_scenario.rs
2. Implement the DemoScenario trait
3. Register in crates/ix-agent/src/demo/scenarios/mod.rs
4. The scenario becomes available in all modes (MCP, CLI, this skill) automatically