// SCAN REPORT

Qontinui MCP Server

64 tools. 22 can modify or destroy data without limits.

3 destructive tools with no built-in limits. Policy required.

Last updated: 4 Jul 2026

22 can modify or destroy data

42 read-only

64 tools total

Community server · catalogue entry verified 04/07/2026

How to control Qontinui MCP Server ↓

TOOL MAP

What Qontinui MCP Server exposes to your agents

Read (42) Write / Execute (19) Destructive / Financial (3)

Running more than one server? Check your whole stack →

MOST DANGEROUS TOOLS

Critical Risk

The most dangerous Qontinui MCP Server tools

Destructive · High delete_test The tool irreversibly deletes data (a verification test) and cannot be reversed. Destructive · Medium mem_delete The tool performs a delete operation on an observation. Destructive · Medium mem_cleanup Archiving under a retention policy typically involves irreversibly removing or moving data (old, low-value observations) from active storage. Execute · High awas_execute This is an Execute-category tool because it runs external operations (HTTP requests to websites) triggered by user-supplied action manifests.

22 of Qontinui MCP Server's 64 tools can modify, destroy, or commit something on every call — and an agent calls them with no built-in limits.

POLICY

How to control Qontinui MCP Server

PolicyLayer is an MCP gateway — it sits between your AI agents and Qontinui MCP Server, and nothing reaches the server without passing your rules. These are the rules we recommend:

Deny destructive operations

{
  "delete_test": {
    "deny_if": [
      {
        "conditions": [],
        "on_deny": "Blocked by default. Requires approval."
      }
    ]
  }
}

Destructive tools should never be available to autonomous agents without human approval.

Rate limit write operations

{
  "build_gui_config": {
    "limits": [
      {
        "counter": "build_gui_config_per_hour",
        "window": "hour",
        "max": 30,
        "scope": "grant"
      }
    ]
  }
}

Prevents bulk unintended modifications from agents caught in loops.

Cap read operations

{
  "awas_check_support": {
    "limits": [
      {
        "counter": "awas_check_support_per_minute",
        "window": "minute",
        "max": 60,
        "scope": "grant"
      }
    ]
  }
}

Controls API costs and prevents retry loops from exhausting upstream rate limits.

Create a free account and register Qontinui MCP Server — nothing to install.
Add these rules — paste them, or build them visually. Tune the limits to your setup.
Point your MCP client (Claude, Cursor, anything) at your gateway URL.

ENFORCE POLICY ON QONTINUI →

Instant setup, no code required.

FULL CATALOGUE

All 64 Qontinui MCP Server tools

DESTRUCTIVE 3 tools

Destructive delete_test Delete a verification test by ID. This is permanent and cannot be undone. Destructive mem_cleanup Run retention policy to archive old, low-value observations. Destructive mem_delete Soft-delete an observation by ID.

EXECUTE 13 tools

Execute awas_execute Execute an AWAS action on a website. The manifest must be discovered first using awas_discover. This tool make Execute capture_multi_state_gui_config Capture a complete multi-state GUI automation config in one call. Execute check_page_spec Run B-style spec verification against the live UI. Loads the page Execute ensure_config_loaded Ensure a specific config file is loaded. Loads it if not already loaded, skips if already loaded. Execute execute_plan Execute a structured implementation plan with sequential AI phases. Each phase runs as a separate AI session w Execute execute_python Execute inline Python code. Returns stdout, stderr, and optional return value. Use this for quick data analysi Execute execute_state_transition Execute a specific transition by ID in the loaded state machine. Execute execute_test Execute a verification test by ID. Returns execution result including pass/fail status, output, assertions, an Execute generate_workflow Generate a UnifiedWorkflow from a natural language description using AI. Execute navigate_to_states Navigate to target states using pathfinding. Execute run_workflow Run a workflow by name from the currently loaded configuration. Execute spawn_sub_agent Spawn a sub-agent with a specific task and optionally scoped tools. The sub-agent runs autonomously and return Execute stop_execution Stop the currently running workflow execution.

WRITE 6 tools

Write build_gui_config Build a visual GUI automation config (QontinuiConfig) from captured Write create_test Create a new verification test. Use this to define automated tests that verify application behavior. Supports Write mem_save Save a persistent observation (decision, architecture, bugfix, pattern, learning, discovery). Write mem_update Update an existing observation Write migrate_task_run_logs Migrate JSONL logs to SQLite for a task run. This persists the current .dev-logs/.jsonl files to the database Write update_test Update an existing verification test by ID. Only provided fields will be updated.

READ 42 tools

Read awas_check_support Check if a website supports AWAS (AI Web Action Standard). Returns a summary of the website Read awas_discover Discover AWAS manifest for a website. AWAS (AI Web Action Standard) enables AI agents to interact with web app Read awas_list_actions List available AWAS actions for a website. Returns the action definitions including their parameters, HTTP met Read capture_gui_elements Capture element images from the runner Read describe_page_spec Read the bundled-page projection of a single spec, identified by Read get_active_states Get the currently active states in the loaded state machine. Read get_annotated_screenshot Get an annotated screenshot with element IDs and bounding boxes for AI consumption. Read get_automation_run Get a specific automation run with full details including actions_summary, states_visited, transitions_execute Read get_automation_runs Get recent automation runs from the runner database. These are GUI automation workflow executions with detaile Read get_available_transitions Get all transitions that are currently available from the active states. Read get_dom_capture Get metadata for a specific DOM capture by ID. Read get_dom_capture_html Get the full HTML content of a DOM capture. Use this to analyze page structure for debugging UI/styling issues Read get_executor_status Get the current status of the Qontinui Runner. Read get_interaction_heatmap Get an interaction heatmap showing clickable/interactive regions. Read get_loaded_config Get information about the currently loaded configuration, including available workflows. Read get_state_machine_status Get the status and statistics of the currently loaded state machine. Read get_task_run Get a specific task run with full details including execution_steps_json and output_log. Read get_task_run_events Get events for a task run from SQLite database. Use this for historical queries on past task runs. For real-ti Read get_task_run_playwright_results Get Playwright test results for a task run from SQLite database. Returns test outcomes, durations, errors, and Read get_task_run_screenshots Get screenshots for a task run from SQLite database. Returns screenshot metadata including file paths, templat Read get_task_runs Get all task runs from the runner database. Optionally filter by status. Read get_test Get a specific verification test by ID with full details including code and configuration. Read get_test_history Get test history summary with aggregated statistics including pass rate, total runs, and recent results. Read get_visual_diff Get a visual diff highlighting changes between two screenshots. Read list_dom_captures List DOM captures (HTML snapshots) from browser pages. Use this to find captured page HTML for debugging UI/st Read list_monitors List available monitors with position information (left, middle, right, primary). Read list_page_specs Enumerate every page spec registered under Read list_screenshots List available screenshots in the .dev-logs/screenshots directory. Returns file paths that can be read with Cl Read list_test_results List test results with optional filtering by test, task run, or status. Read list_tests List all verification tests. Filter by type (playwright_cdp, qontinui_vision, python_script, repository_test) Read load_config Load a JSON workflow configuration file into the runner. Use this before running workflows. Read load_state_machine Load a UI Bridge state machine configuration into the runner. Read mem_by_task_run Get all observations linked to a specific task run. Read mem_context Get relevant observations for a project. Returns recent observations Read mem_export Export all observations as JSON for backup or sharing. Read mem_get Get the full content of an observation by ID. Use after mem_search Read mem_search Full-text search over persistent observations. Returns 300-char previews Read mem_stats Get observation statistics grouped by type. Read project_current_scenario Get a runtime-aware scenario projection - like project_scenarios, but Read project_scenarios Get a deterministic static scenario projection of an IR document - Read read_runner_logs Read runner JSONL log files from .dev-logs directory. Use this to access detailed execution logs including ima Read validate_page_spec Validate the on-disk IR for a page spec against the G2

EXPLORE

Related servers

Other MCP servers with similar tools — same risk classification, starter policies for each.

FAQ

Questions about Qontinui MCP Server

Can an AI agent delete data through the Qontinui MCP Server MCP server? +

Yes. The Qontinui MCP Server server exposes 3 destructive tools including delete_test, mem_cleanup, mem_delete. These permanently remove resources with no undo. PolicyLayer blocks destructive tools by default so they never reach the upstream server.

How do I prevent bulk modifications through Qontinui MCP Server? +

The Qontinui MCP Server server has 6 write tools including build_gui_config, create_test, mem_save. Set a rate limit in your policy -- for example, 10 calls per hour prevents an agent from making more than 10 modifications per hour. PolicyLayer enforces this at the gateway, before calls reach Qontinui MCP Server.

How many tools does the Qontinui MCP Server MCP server expose? +

64 tools across 4 categories: Destructive, Execute, Read, Write. 42 are read-only. 22 can modify, create, or delete data.

How do I enforce a policy on Qontinui MCP Server? +

Register the Qontinui MCP Server MCP server in PolicyLayer, apply the suggested rules above (adjust the limits to your use case), and point your AI client at the PolicyLayer proxy URL instead of the server directly. Your agents keep the same tools; PolicyLayer evaluates every call against policy before it executes. Nothing to install, live in minutes.

Enforce policy on every Qontinui MCP Server tool call.

Deterministic rules across all 64 Qontinui MCP Server tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.

ENFORCE POLICY ON QONTINUI →

Instant setup, no code required.

64 Qontinui MCP Server tools catalogued and risk-classified — across an index of 43,000+ MCP servers.