Skip to main content
Glama
audatisControl

audatisControl

by Truecrypt313

audatisControl

Unofficial MCP server for AI-assisted audatis MANAGER administration

License: PolyForm Noncommercial 1.0.0 · Commercial licensing · First release notes

audatisControl is a Python Model Context Protocol (MCP) server that connects MCP-compatible AI clients (such as GitHub Copilot in VS Code) to the audatis MANAGER REST API. Ask your assistant to inspect assets, review documentation quality, manage supported records, and carry out controlled bulk operations.

Independent project: This repository is not affiliated with, sponsored by, or endorsed by audatis. audatis MANAGER and related trademarks belong to their respective owners. This repository does not distribute the vendor's API specification.

Features

Area

Capabilities

IT assets

List, create, update, owner assignments, protection overrides, supported parent/child hierarchy

Tasks

List, create, update, prioritise, inspect potential duplicates

Information values

Read and manage supported information records

Business processes

Read, create, update, inspect and manage parent/child hierarchy

Security measures

Read, create, update, assess documented coverage

Risks

Read and analyse available risk records; no risk-writing endpoint implemented

Incidents

Read incidents and manage supported links to existing records; no incident creation

Personnel

Read, create, update personnel records; assign them to existing organisations

Organisations and users

Read existing organisation units and administrative users; no account or organisation creation

Analysis

Inventory snapshots, documentation gaps, relationship graph and audit

Bulk operations

Preview, validate and execute selected batched create/update operations

47 MCP tools are currently registered. Set AUDATIS_MCP_TOOL_NAMES=modern for consistent public names without the older v2_ and v21_ prefixes. The original names remain available with legacy mode for compatibility.

Example prompts

These examples assume that the MCP server is connected to your AI client:

  • Use list_assets and documentation_gaps to find incomplete asset documentation. Do not modify anything.

  • Use relationship_graph and related_records to show the documented dependencies of this asset.

  • Use list_personnel and list_organisations to show personnel assignments. Read-only.

  • Preview creating two test tasks via bulk_create_records with execute=false. Do not execute writes.

  • Use manager_capabilities and lookup_values to determine valid fields and status codes before preparing updates.

The AI client may select the underlying tool based on natural-language instructions; these names are MCP tools, not shell commands.

Related MCP server: Bullhorn MCP Agent

Architecture

MCP-compatible AI client (e.g. VS Code / GitHub Copilot)
             |
           stdio
             |
    audatisControl (Python MCP server)
      |     |     |
    Tools  Validation  Tenant guard
             |
       audatis MANAGER REST API

The integration does not perform automatic network discovery. Graphs represent relationships already documented in audatis MANAGER.

Quick start

Requirements: Python 3.11+, a permitted audatis MANAGER API endpoint, and valid API credentials.

git clone https://github.com/Truecrypt313/audatis-control-mcp.git
cd audatis-control-mcp
python3 -m venv .venv
source .venv/bin/activate
python -m pip install -r requirements.txt -r requirements-v2.txt
cp .env.example .env

Edit .env locally:

AUDATIS_BASE_URL=https://your-audatis-host.example
AUDATIS_CLIENT_ID=12345
AUDATIS_ALLOWED_CLIENT_ID=12345
AUDATIS_API_TOKEN=YOUR_READ_ONLY_TOKEN
AUDATIS_MCP_TOOL_NAMES=modern
  • AUDATIS_CLIENT_ID selects the tenant.

  • AUDATIS_ALLOWED_CLIENT_ID explicitly scopes supported write operations. Never point writable credentials at a production tenant without appropriate controls.

  • AUDATIS_API_TOKEN is the read API credential.

  • Store a separate write token, if needed, in ~/.config/audatis-mcp/write_token with restrictive permissions:

mkdir -p ~/.config/audatis-mcp
chmod 700 ~/.config/audatis-mcp
# Add your write token using a local editor; do not paste it into chat or Git.
chmod 600 ~/.config/audatis-mcp/write_token

Use .vscode/mcp.example.json as a template for your local .vscode/mcp.json. Replace the placeholder absolute Python and server.py paths. Keep the example and real configuration separate; never commit a machine-specific configuration containing secrets.

For direct stdio execution:

.venv/bin/python server.py

A stdio MCP server communicates through its protocol; it does not normally print a human-readable application menu in this mode. Start it through your configured MCP client.

Testing

source .venv/bin/activate
python -m pytest -q tests/

The offline test suite uses simulated API interactions. It cannot confirm your server's actual permissions or live audatis MANAGER compatibility. Test live writes only in a dedicated, authorised tenant.

Security & operational limitations

  • Real writes are possible: Creating or changing records can affect persistent data. Preview first when supported.

  • Tenant scope: Configure both the active and permitted write-tenant IDs. Use narrowly scoped credentials; confirm each legacy/specialised tool's permissions before deployment.

  • No transactional bulk rollback: An individual write can succeed even if a subsequent write fails. Verify remote state before attempting retries.

  • No arbitrary API proxy: Only supported resource types and fields are exposed by the tools.

  • Privacy: Never commit credentials, real customer data, internal hostnames, or local MCP configuration.

  • Not a compliance verdict: Documentation-gap and relationship findings are heuristic indicators, not an audit certification or a substitute for professional security assessment.

  • API coverage: Users cannot create administrative accounts or organisation units through this implementation; risks are read-only; incident functionality is limited to reading and supported links.

License and commercial use

The software is source-available, not OSI-approved open source:

  • Noncommercial use: Free under the PolyForm Noncommercial License 1.0.0, including permitted personal, hobby, educational, and other noncommercial uses as defined by that license.

  • Commercial use: Requires a separately negotiated paid license. See Commercial Licensing.

  • Third-party dependencies and the audatis MANAGER service retain their own licenses and conditions.

Copyright © 2026 Truecrypt313. See NOTICE.

Project status

Early public release. Contributions, feedback, and bug reports are welcome through GitHub Issues. Please report security-sensitive findings privately rather than posting secrets or exploit details in public issues.

Related MCP Connectors

Related MCP Servers

  • F
    license
    B
    quality
    B
    maintenance
    Enables MCP-compatible AI clients to securely search and retrieve Bullhorn jobs, candidates, and placements, manage candidate files and resumes, and run generic entity searches/queries without exposing credentials.
    23
    -
  • A
    license
    Not graded
    quality
    B
    maintenance
    Enables browser-based AI services to connect through remote MCP for internal document search, email checking and replies, timecard, sales/project, schedule/task, and operations tasks. It supports multiple AI clients via compatible remote MCP endpoints.
    Apache 2.0
  • A
    license
    Not graded
    quality
    B
    maintenance
    Exposes scattered business REST APIs and read-only databases to AI clients through a single MCP endpoint with unified API-key authentication, service- and operation-level authorization, department-scoped filtering, masking, and audit logging. AI agents discover only the tools and data they are permitted to use, without needing to know endpoint addresses, SQL, or credentials.
    MIT