Mason Context MCP Server

Tools that read sensitive data ([get_code_samples]) and tools that can send data out ([export_to_confluence]) are exposed together. An agent can move private data to the sink.

Evidence: sources [get_code_samples] → sinks [export_to_confluence]

Fix: Keep secret-reading and egress capabilities on separate, separately-approved servers.

Location: flow get_code_samples → export_to_confluence

In the server's implementation (`dist/mason-auto.js:48`): Reading private keys / cloud credentials, or serializing the whole environment, is a sensitive-data source that becomes exfiltration when combined with any egress. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: => /^(?:\.env(?:\..*)?|id_rsa.*|id_ed25519.*)$|\.(?:pem|key|p12|pfx|jks|keystore)$|credentials\.|secret|^local\.propert

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server dist/mason-auto.js

In the server's implementation (`dist/mason-drift.js:77`): Reading private keys / cloud credentials, or serializing the whole environment, is a sensitive-data source that becomes exfiltration when combined with any egress. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: => /^(?:\.env(?:\..*)?|id_rsa.*|id_ed25519.*)$|\.(?:pem|key|p12|pfx|jks|keystore)$|credentials\.|secret|^local\.propert

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server dist/mason-drift.js

In the server's implementation (`dist/mason-mcp.js:271`): Reading private keys / cloud credentials, or serializing the whole environment, is a sensitive-data source that becomes exfiltration when combined with any egress. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: => /^(?:\.env(?:\..*)?|id_rsa.*|id_ed25519.*)$|\.(?:pem|key|p12|pfx|jks|keystore)$|credentials\.|secret|^local\.propert

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server dist/mason-mcp.js

In the server's implementation (`dist/mason-review.js:84`): Reading private keys / cloud credentials, or serializing the whole environment, is a sensitive-data source that becomes exfiltration when combined with any egress. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: => /^(?:\.env(?:\..*)?|id_rsa.*|id_ed25519.*)$|\.(?:pem|key|p12|pfx|jks|keystore)$|credentials\.|secret|^local\.propert

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server dist/mason-review.js

In the server's implementation (`dist/mason.js:48`): Reading private keys / cloud credentials, or serializing the whole environment, is a sensitive-data source that becomes exfiltration when combined with any egress. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: => /^(?:\.env(?:\..*)?|id_rsa.*|id_ed25519.*)$|\.(?:pem|key|p12|pfx|jks|keystore)$|credentials\.|secret|^local\.propert

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server dist/mason.js

In the server's implementation (`dist/mason-audit.js:13`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: le as execFile3 } from "child_process"; import { promisify as promisify3 } from "util"; // src/snapshot/snapshot.ts imp

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server dist/mason-audit.js

In the server's implementation (`dist/mason-auto.js:43`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: port { execFile } from "child_process"; import { promisify } from "util"; import fg from "fast-glob"; function isSensiti

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server dist/mason-auto.js

In the server's implementation (`dist/mason-drift.js:9`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: le as execFile3 } from "child_process"; import { promisify as promisify3 } from "util"; // src/snapshot/snapshot.ts imp

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server dist/mason-drift.js

In the server's implementation (`dist/mason-hook.js:40`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: port { execFile } from "child_process"; import { promisify } from "util"; import fg from "fast-glob"; var exec = promisi

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server dist/mason-hook.js

In the server's implementation (`dist/mason-mcp.js:266`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: port { execFile } from "child_process"; import { promisify } from "util"; import fg from "fast-glob"; function isSensiti

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server dist/mason-mcp.js

In the server's implementation (`dist/mason-review.js:10`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: le as execFile5 } from "child_process"; import { promisify as promisify5 } from "util"; // src/drift/drift.ts import fs

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server dist/mason-review.js

In the server's implementation (`dist/mason.js:43`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: port { execFile } from "child_process"; import { promisify } from "util"; import fg from "fast-glob"; function isSensiti

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server dist/mason.js

In the server's implementation (`dist/mason.js:5396`): Loading a module chosen at runtime (from a variable) can pull in and run attacker-influenced code paths. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: y), ...args]; await import(binary.href); } else throw new Error("Unknown command: " + command + ". Run mason --hel

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server dist/mason.js