[Feature / Hacktoberfest] Expand WebMCP Agentic Simulations for Student Competency Proofs

Context & Overview

In branch agentic-evaluation, we introduced an experimental implementation of Google WebMCP (Web Model Context Protocol) in sites/student-site.

The Autonomy Protocol transitions high school students from external behavioral policing to internal self-governance. Rather than relying on static multiple-choice questionnaires, WebMCP turns student assessments into interactive, agent-driven practical simulations. An autonomous evaluator agent (running locally in-page, via the Chrome Model Context Tool Inspector Extension, or through browser agents) communicates with the student's tab using standardized WebMCP tools (document.modelContext) to audit cognitive liabilities, verify restorative restitution, and issue verifiable proofs of competency.

What's Built So Far

1. WebMCP Client Core & Lifecycle (src/lib/webmcp/client.ts):

   • Primary binding to document.modelContext with backward fallback to navigator.modelContext.
   • Dynamic and static Origin Trial token registration configured for the-autonomy-protocol-student.vercel.app.
   • Signal-based lifecycle teardown via AbortController.abort() to prevent memory leaks.
   • Built-in fallback runtime so simulations run 100% reliably in any browser.

2. The Accountability Cycle Simulation Engine (src/lib/webmcp/simulationEngine.svelte.ts):

   • Svelte 5 runes ($state) modeling the 4 core phases:
     • Orientation (Steps 1–3): Limits of control, objective standard committal.
     • Internal Audit (Steps 4–7): Identification & refactoring of cognitive distortions (Externalized Blame, Catastrophizing, Assumptive Projection).
     • Restorative Practice (Steps 8–9): Impact mapping and action-based Amends Protocol.
     • Maintenance & Leadership (Steps 10–12): Conflict de-escalation under environmental friction.
   • 6 registered atomic WebMCP tools:
     • get_student_competency_profile
     • inspect_state_machine
     • inject_environmental_friction
     • audit_cognitive_alignment
     • verify_amends_protocol
     • record_proof_of_competency

3. Collegiate Desktop UI Workbench (/simulation):

   • Real-time Status Banner with live tool counter & token configuration.
   • Accountability Cycle Pipeline Visualizer tracking active state nodes.
   • Student Interactive Control Deck to input refactorings and restorative plans.
   • Autonomous Evaluator Agent Harness to trigger proctor runs and view signed proof-of-competency certificates.
   • Agent Telemetry Stream showing real-time JSON payloads and tool executions.

How You Can Contribute (Open Opportunities)

We welcome pull requests from developers, Hacktoberfest participants, and educational strategists! Here are high-impact areas ready for extension:

1. Industry-Specific Capstone Pathways (Beginner / Intermediate)

Currently, the simulation defaults to a Computer Science sprint delivery scenario.

• Add scenario templates for Healthcare / Nursing (e.g., patient boundary triage, charting dispute de-escalation).
• Add scenario templates for Construction / Skilled Trades (e.g., subcontractor supply delays, safety liability audits).
• Implement a pathway switcher dropdown on /simulation.

2. IndexedDB / Local Persistence for Competency Proofs (Intermediate)

• Persist CompetencyVerificationRecord objects in browser IndexedDB or localStorage.
• Wire the records into the student portal's existing database / achievements store (src/lib/stores/moduleStore.svelte.ts) so credentials persist across sessions.

3. Declarative WebMCP Form Annotations (Intermediate)

Chromium WebMCP supports both Imperative JavaScript tools and Declarative HTML annotations:

• Add declarative HTML form annotations (toolname, tooldescription, toolparameter) to the student input forms as a secondary registration surface.

4. Interactive Integrity Protocol Calibrator (Intermediate)

• Build interactive calibration sliders or linguistic diff highlighters for the 4 Integrity Protocol tenets:
  • Precision of Speech
  • Emotional Neutrality
  • Inquiry-Based Reality
  • Baseline Fluctuation
• Expose an additional WebMCP tool evaluate_speech_precision that audits student text for subjective adjectives vs. objective telemetry metrics.

5. Prompt Templates & Inspector Integration (Beginner)

• Author a set of recommended evaluator agent prompts for the Chrome Model Context Tool Inspector Extension.
• Add a "Copy Suggested Agent Prompts" drawer to the UI.

Join the WebMCP origin trial from Chrome 149.

• Sign in and register for the trial
• Register domain Student Site
• Add the token to the .env file.

Getting Started Locally

You will need Vite+
https://viteplus.dev/guide/

curl -fsSL https://vite.plus | bash

Repo Setup

# 1. Clone & checkout the branch
git checkout agentic-evaluation

# 2. Enter student dir
cd sites/student-site

# 2. Install monorepo dependencies
vp i

# 3. Launch the student site dev server
`vpr dev`

# 4. Open the simulation workbench
`open http://localhost:6173/simulation`

Optional: Enable Native Chrome WebMCP Flags

• Navigate to chrome://flags/#enable-webmcp-testing in Chromium 146+.
• Set to Enabled and relaunch to test native browser actuation without polyfills.

Contribution Guidelines

• Follow the pedagogical constraints in GEMINI.md: Maintain academic, professional, non-secular framing (focus on observable behavioral metrics rather than subjective emotional states).
• Follow Svelte 5 Runes standards ($state, $derived, $props).
• Test locally with pnpm --filter student-site check and pnpm --filter student-site build.

Choose one of the independently listed opportunities before starting. Read the relevant entry point in sites/student-site, such as src/lib/webmcp/client.ts, src/lib/webmcp/simulationEngine.svelte.ts, src/lib/stores/moduleStore.svelte.ts, or the /simulation UI, then run pnpm --filter student-site check and build. Done means the selected pathway, persistence, annotation, calibration, prompt, or origin-trial task works in the student site and both checks pass.