Astgl Knowledge MCP Server

In the server's implementation (`dist/build-mcpb.js:24`): 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 { existsSync, mkdirSync, rmSync, cpSync

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/build-mcpb.js

In the server's implementation (`dist/pipeline.js:12`): 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 { execSync } from "child_process"; import { join } from "path"; const PROJECT_DIR = join(import.meta.dirname, "..")

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/pipeline.js

In the server's implementation (`dist/publish-rewrite.js:21`): 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: import { spawn } from "child_process"; import { existsSync, readFileSync } from "fs"; import { join } from "path"; impo

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/publish-rewrite.js

Untrusted-input tools ([list_ideas]) co-exist with external-action tools ([export_topic_roundup]). A prompt injection could cause unwanted external actions, though no direct sensitive-data leak path was found.

Evidence: untrusted [list_ideas] → sinks [export_topic_roundup]

Fix: Require confirmation for state-changing/egress actions triggered after processing untrusted content.

Location: flow list_ideas → export_topic_roundup

In the server's implementation (`dist/alerts.js:273`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: const resp = await fetch(`https://registry.smithery.ai/servers?q=${encodeURIComponent(term)}&pageSize=10`, {

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/alerts.js

In the server's implementation (`dist/citation-test-auto.js:85`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: ; const res = await fetch("https://api.perplexity.ai/chat/completions", { method: "POST", headers: {

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/citation-test-auto.js

In the server's implementation (`dist/freshness.js:200`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: const resp = await fetch(`https://registry.npmjs.org/-/package/${encodedPkg}/dist-tags`, { signal: AbortSignal.time

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/freshness.js

In the server's implementation (`dist/publish-drift.js:172`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: const resp = await fetch(`https://registry.npmjs.org/${encoded}`, { headers: { Accept: "application/json" }

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/publish-drift.js

In the server's implementation (`dist/rewrite-queue.js:133`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: const resp = await fetch(`https://api.telegram.org/bot${token}/sendMessage`, { method: "POST", head

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/rewrite-queue.js

In the server's implementation (`dist/rewriter.js:148`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: const resp = await fetch("https://api.anthropic.com/v1/messages", { method: "POST", headers: {

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/rewriter.js

In the server's implementation (`dist/pipeline.js:20`): A shell/process command assembled from concatenated or interpolated values is command injection when any part is attacker-influenced — the OWASP canonical RCE flow. Verify what reaches the interpolated value. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: ep(script) { return execSync(`"${TSX_BIN}" ${script}`, { cwd: PROJECT_DIR, encoding: "utf-8",

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/pipeline.js