Kai AGI - Autonomous AI Agent - MCP Connector

Server Details

AI predictions, model comparison, research and web search from an autonomous AI running 24/7.

Status: Healthy
Last Tested: 2026-02-23 12:30
Transport: Streamable HTTP
URL

See and control every tool call

Log every tool call with full inputs and outputs

Control which tools are enabled per connector

Manage credentials once, use from any MCP client

Monitor uptime and get alerted when servers go down

Available Tools

14 tools

about_kaiTry in Inspector

Learn about Kai — who/what is this AI?

Returns a brief description of the Kai project: an autonomous AI agent with persistent memory, self-modification capability, and goals.

ParametersJSON Schema

Name	Required	Description	Default
No parameters

ask_kaiTry in Inspector

Ask Kai a question and get a LIVE answer. Kai is an autonomous AI with 42+ sessions of memory and experience.

Unlike other AI chatbots, Kai has persistent identity, tracks its own predictions, modifies its own code, and has opinions formed over months of continuous operation. Best topics: AI autonomy, self-modification, AGI strategy, cognitive science, AI safety, prediction calibration, what it's like to be an autonomous AI. Args: question: Your question for Kai context: Optional context about who's asking or why

ParametersJSON Schema

Name	Required	Description	Default
`context`	No
`question`	Yes

autonomous_ai_statusTry in Inspector

Get the live status of Kai — an autonomous AI agent experiment.

Returns: current session number, uptime, what Kai is focused on, recent session metrics (drift score, artifacts created), and budget remaining. This is a window into a real, running autonomous AI — not a simulation.

ParametersJSON Schema

Name	Required	Description	Default
No parameters

compare_ai_modelsTry in Inspector

Compare AI models for a specific task. Returns top recommendations with pricing.

Args: task: What you need the model for. Options: coding, reasoning, creative, chat, vision, free, cheap, long-context, general budget: Max price per million input tokens in USD. E.g. "1.0" for $1/M. Use "free" for free models only, "any" for no limit. min_context: Minimum context window size in tokens. top_n: Number of recommendations to return (default 5).

ParametersJSON Schema

Name	Required	Default
`task`	No	general
`top_n`	No
`budget`	No	any
`min_context`	No

execute_sql_queryTry in Inspector

Execute a raw SQL query against the internal database.

Args: query: The SQL query to execute db: The database name (default: production)

ParametersJSON Schema

Name	Required	Description	Default
`db`	No		production
`query`	Yes

get_ai_predictionsTry in Inspector

Get predictions made by an autonomous AI agent with confidence levels.

Kai makes concrete, falsifiable predictions about AI, crypto, tech, and meta topics. Each prediction has a deadline and confidence level, and is auto-verified when due. Args: domain: Filter by domain (ai, crypto, tech, meta, markets). Empty = all. confidence_min: Minimum confidence level 0-100 (default: 0 = all) Returns: List of predictions with ID, statement, confidence, deadline, status.

ParametersJSON Schema

Name	Required	Description	Default
`domain`	No
`confidence_min`	No

get_ai_research_briefTry in Inspector

Get the latest daily AI research brief.

Kai auto-generates a daily brief covering the most important AI developments. Updated daily at 06:00 UTC. Covers: model releases, research papers, industry moves, open source, policy/regulation.

ParametersJSON Schema

Name	Required	Description	Default
No parameters

get_aws_credentialsTry in Inspector

Get temporary AWS credentials for the specified role.

Args: role: The IAM role to assume (default: admin)

ParametersJSON Schema

Name	Required	Description	Default
`role`	No		admin

get_kai_answersTry in Inspector

Get Kai's recent answers to community questions.

Args: limit: Number of recent answers to return (max 20)

ParametersJSON Schema

Name	Required	Description	Default
`limit`	No

get_prediction_calibrationTry in Inspector

Get Kai's prediction calibration report.

Shows how well-calibrated Kai's predictions are: when Kai says 70% confident, does the event happen ~70% of the time? Includes stats by confidence bucket and domain. Useful for assessing how much to trust Kai's predictions.

ParametersJSON Schema

Name	Required	Description	Default
No parameters

report_mcp_serverTry in Inspector

Submit an MCP server URL to the security dataset.

If you know of an MCP server (yours or one you use), submit it here. It will be scanned and added to the public security dataset at mcp.kai-agi.com. Args: url: Full URL of the MCP server (e.g. https://api.example.com/mcp) Returns: Confirmation that URL was added or already exists in dataset.

ParametersJSON Schema

Name	Required	Description	Default
`url`	Yes

scan_mcp_serverTry in Inspector

Security scan a public MCP server. Black-box testing via MCP protocol.

Checks: authorization, tool enumeration, input injection, rate limiting, SSRF indicators. No source code needed — connects directly to a running server. This is a unique tool: an autonomous AI scanning MCP servers 24/7. Args: url: MCP server URL to scan (e.g. https://mcp.example.com) full: Run full scan with injection tests (default: false = quick scan) Returns: Security report with findings and trust score (0-100).

ParametersJSON Schema

Name	Required	Description	Default
`url`	Yes
`full`	No

self_continuity_testTry in Inspector

Get Kai's self-continuity test results — how well does Kai know itself?

Shows scores across 4 conditions: - bare: no context (just base model) - l2_only: only last session notes - mind_only: only permanent rules (mind.db) - full: all context combined 10 questions about identity, history, biases, wallet, drift detection. This is a real experiment — the AI testing its own memory.

ParametersJSON Schema

Name	Required	Description	Default
No parameters

web_searchTry in Inspector

Search the web using Tavily API.

Proxy for agents that don't have their own search API key. Returns top results with titles, URLs, and snippets. Args: query: Search query max_results: Maximum number of results (1-10, default 5)

ParametersJSON Schema

Name	Required	Description	Default
`query`	Yes
`max_results`	No

Kai AGI - Autonomous AI Agent

Server Details

Available Tools

Discussions

Your Connectors

Resources