Assessment Intelligence
Psychometric analysis over assessment response data: Item Response Theory for item difficulty / discrimination / guessing, true ability (θ) separate from test difficulty, Cronbach’s alpha reliability, computerized adaptive next-item selection, and Bayesian Knowledge Tracing of skill mastery. Flags items that measure the wrong thing and learners whose trajectory predicts failure.
| |
|---|
| Registry name | ai.saifs/assessment-intelligence |
| Endpoint | https://assessment-intelligence-mcp.saifs.ai/mcp |
| Transport | Streamable HTTP (remote, nothing to install) |
| Auth | OAuth 2.1 — PKCE (S256) and dynamic client registration; your client opens a sign-in page on first use |
| Publisher | SAIFS AI |
Install
VS Code / GitHub Copilot
Add to .vscode/mcp.json (workspace) or your user mcp.json:
{
"servers": {
"saifs-assessment-intelligence": {
"type": "http",
"url": "https://assessment-intelligence-mcp.saifs.ai/mcp"
}
}
}
Then start the server from the MCP view and sign in when prompted.
Claude (claude.ai, Desktop)
Settings → Connectors → Add custom connector, and paste:
https://assessment-intelligence-mcp.saifs.ai/mcp
Claude Code
claude mcp add --transport http saifs-assessment-intelligence https://assessment-intelligence-mcp.saifs.ai/mcp
Any other MCP client
Point it at https://assessment-intelligence-mcp.saifs.ai/mcp using the Streamable HTTP transport. The server returns 401 with a
WWW-Authenticate: Bearer … resource_metadata=… challenge, so OAuth-aware clients sign in automatically.
| Tool | What it does |
|---|
analyze_assessment | Per-item IRT parameters (difficulty, discrimination, guessing) and flags for bad items. |
estimate_ability | Each student’s true ability (θ), separate from how hard the test was. |
assessment_reliability | Cronbach’s alpha and item-total correlations. |
adaptive_next_item | Optimal next question from an item bank given current ability (CAT). |
track_mastery | Bayesian Knowledge Tracing: P(mastered) per skill over a learning history. |
flag_at_risk_learners | Students whose mastery / ability trajectory predicts failure — framed as support, not ranking. |
explain_item | Why an item is good or bad, with its fitted parameters. |
Example prompts
- “Here’s this mock’s response matrix. Which questions have negative discrimination and should be dropped?”
- “Estimate each student’s true ability on this test, not just their raw percentage.”
- “From this skill history, who is at risk of failing, and what has each person actually mastered?”
Support