discourse/discourse-mcp — agentic threat model
The discourse-mcp agent exposes read and write forum operations, presenting a moderate risk profile where user-generated forum content can act as an indirect prompt injection vector to abuse administrative API keys.
OWASP AIVSS score rationale
| Autonomy of Action | 0.40 | |
| Goal-Driven Planning | 0.30 | |
| Self-Modification | 0.00 | |
| Dynamic Tool Use | 0.60 | |
| Persistent Memory | 0.10 | |
| Contextual Awareness | 0.50 | |
| Dynamic Identity | 0.40 | |
| Multi-Agent Interactions | 0.20 | |
| Non-Determinism | 0.50 | |
| Opacity & Reflexivity | 0.30 |
Scored with the canonical OWASP AIVSS formula (AIVSS calculator reference); agentic risk factors estimated from the agent’s described capabilities.
MAESTRO 7-layer threat model
Per-layer threats for this agent. Layers tagged “not certain from listing” are general, caveated commentary where the public description didn’t pin that layer.
Not certain from the listing — the MCP server itself is model-agnostic, but the host LLM is vulnerable to indirect prompt injection via user-generated forum posts and topic content retrieved during search operations.
The agent reads directly from Discourse forum databases/APIs. The primary threat is data poisoning or indirect prompt injection embedded in forum posts, categories, or tags, which are then ingested by the agent.
Exposes tools for topic/post search, category/tag management, and user discovery. Vulnerable to tool misuse if the orchestrating framework executes write operations (like category management) without strict validation or human-in-the-loop approval.
Not certain from the listing — depends on how the MCP host is deployed, but the server requires access to Discourse API keys. If these keys are stored insecurely in the environment, it could lead to credential theft and full forum compromise.
Not certain from the listing — there is no mention of built-in logging, audit trails, or guardrails to monitor the agent's tool execution or detect anomalous administrative actions on the forum.
The listing notes that API keys scope moderation abilities. A major risk is privilege escalation or unauthorized administrative actions if the API key used by the MCP server has excessive permissions (e.g., admin/moderator bypass).
As an MCP tool, this agent can be integrated into multi-agent workflows. A compromised or malicious orchestrator agent could abuse the Discourse tools to spam, delete topics, or exfiltrate private community data.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).