MCP and agent packages
The packages that advisories about the Model Context Protocol and AI agents name, with what the Exposure Registry knows about each one: the LLM components it declares and the authority its dependencies grant (shell, files, browser, network, code execution).
Advisories are grouped by topic. The Model Context Protocol view counts the records that match that topic, and the AI agents view counts those that match the AI agents or the agent-to-agent protocols topic. Package names are those the advisory source lists. How authority is measured
- Advisories
- 198
- Critical or high
- 148
- Packages named
- 77
- Advisories listed as exploited (CISA KEV)
- 2
99 advisories were published in the last 90 days and 50 in the 90 days before. 64 of the 198 name no package, because their source lists none.
Advisories by month of publication
| Month | Items |
|---|---|
| May 2025 | 1 |
| Jun 2025 | 2 |
| Jul 2025 | 4 |
| Aug 2025 | 1 |
| Sep 2025 | 2 |
| Oct 2025 | 3 |
| Nov 2025 | 0 |
| Dec 2025 | 4 |
| Jan 2026 | 4 |
| Feb 2026 | 5 |
| Mar 2026 | 18 |
| Apr 2026 | 16 |
| May 2026 | 19 |
| Jun 2026 | 13 |
| Jul 2026 | 28 |
| Aug 2026 | 36 |
| Sep 2026 | 31 |
| Oct 2026 | 11 |
Latest advisories
- HighGHSA-hc5v-gxvj-58wh: PraisonAI: MCP HTTP-stream transport is unauthenticated by default, exposing tool enumeration and an unvalidated tool-call surface2026-10-08
- HighGHSA-6qxp-vccf-f47h: MCP TypeScript SDK: OAuth client could send credentials to an authorization server chosen by the MCP server2026-10-06
- HighCVE-2026-105741: Langflow IP spoofing bypasses local-only restriction on MCP install endpoint2026-10-05
- CriticalCVE-2026-105740: Langflow remote code execution through MCP server stdio command field2026-10-05
- HighCVE-2026-105699: Langflow MCP resource read exposes other users' flow files2026-10-05
- CriticalCVE-2026-105697: Langflow arbitrary command execution through MCP stdio server configuration2026-10-05
- MediumGHSA-p23f-cm6q-2qp8: SiYuan MCP asset.upload Reads Arbitrary Absolute File Paths (Workspace Boundary Bypass)2026-10-02
- MediumGHSA-c9xm-49cp-xcr9: rmcp OAuth client fetches server-controlled resource_metadata URLs2026-10-02
Authority in the registry
81 registry packages declare the MCP SDK or FastMCP. Their dependencies grant:
- MCP tools74Exposes or calls Model Context Protocol tools, which pass authority between agents and servers.
- Outbound HTTP38Makes outbound requests, the precondition for server-side request forgery and exfiltration.
- File system9Reads or writes files, so path traversal and data exposure are in reach.
- Browser control7Drives a browser, so it can act on websites with the user's sessions.
- Code execution7Runs code it is given, so injected instructions can become arbitrary code.
- Shell commands3Starts processes on the host, the most direct path from a prompt to the operating system.
Advisories that name mcp-memory-keeper
Closenpm. Every record that names the package, on any topic, newest first. RSS feed for this package
Packages named in advisories
77 packages
| Package | Advisories | Highest severity | Latest advisory | Exploited | Authority |
|---|---|---|---|---|---|
| stata-mcpPyPI | 1 | High | Not in the registry | ||
| gemini-bridgePyPI | 1 | Medium | Not in the registry | ||
| mcp-memory-keepernpm | 1 | Medium | Not in the registry | ||
| langbotPyPI | 1 | High | Not in the registry | ||
| phantom-audioPyPI | 1 | High | Not in the registry | ||
| github.com/coder/coder/v2Go | 1 | Medium | Not in the registry | ||
| @grackle-ai/authnpm | 1 | High | Not in the registry | ||
| @grackle-ai/plugin-corenpm | 1 | High | Not in the registry | ||
| agentic-flownpm | 1 | High | Not in the registry | ||
| anthropics/claude-code-actionactions | 1 | Medium | Not in the registry | ||
| @yoda.digital/gitlab-mcp-servernpm | 1 | Critical | Not in the registry | ||
| @penpot/mcpnpm | 1 | High | Not in the registry | ||
| 9routernpm | 1 | Critical | Not in the registry | ||
| github.com/envoyproxy/ai-gatewayGo | 1 | Medium | Not in the registry | ||
| auth-fetch-mcpnpm | 1 | High | Not in the registry | ||
| @openai/codexnpm | 1 | High | Not in the registry | ||
| io-modelcontextprotocol-sdk:mcp-coremaven | 1 | High | Not in the registry | ||
| @mobilenext/mobile-mcpnpm | 1 | High | Not in the registry | ||
| adx-mcp-serverPyPI | 1 | High | Not in the registry | ||
| github.com/tencent/weknoraGo | 1 | Medium | Not in the registry | ||
| github.com/agentgateway/agentgatewayGo | 1 | Medium | Not in the registry | ||
| @github/copilotnpm | 1 | High | Not in the registry | ||
| mcp-run-pythonPyPI | 1 | Medium | Not in the registry | ||
| @lobehub/chatnpm | 1 | Medium | Not in the registry | ||
| github.com/1panel-dev/1panel/coreGo | 1 | High | Not in the registry | ||
| @cyanheads/git-mcp-servernpm | 1 | High | Not in the registry | ||
| ios-simulator-mcpnpm | 1 | Medium | Not in the registry |
Topics are assigned by matching a record's title and summary against published patterns, so an advisory that never uses the words is missed and one that mentions them in passing is counted. An advisory that names several packages counts once for each. A package is listed with the ecosystem its advisory source states, and by name alone when the source states none.
Authority is read from the dependencies a package's latest release declares, and it is known only for packages the Exposure Registry tracks. It shows what a package's dependencies can do on the host. It does not show that a model is given that reach. Severity comes from the source advisory when it gives one, otherwise from the classifier. Dates are in UTC.