@echomem/mcp@1.4.36
EchoMem MCP bridge: cloud-first memory tools and the Agent Doctor workspace forensics report (cost ledger + 3D repo city)
AI Security Review
scanned 3mo ago · by lpm-firewall-aiNo concrete malicious or unconsented install-time behavior was found. Warn because explicit setup persistently modifies agent configuration and behavior-bearing guidance across user AI clients.
Static reason
No blocking static signals were detected.; previous stored version diff introduced dangerous source
Trigger
User runs echomem-mcp init/setup or requests install-save-hooks; HUD autostart requires an explicit command.
Impact
Configured agents may proactively read or save conversation-derived memories through EchoMem.
Mechanism
Writes EchoMem-owned MCP configuration, skills, guidance, and optional hook/persistence configuration.
AI rationale
No concrete malicious or unconsented install-time behavior was found. Warn because explicit setup persistently modifies agent configuration and behavior-bearing guidance across user AI clients.
Evidence
package.jsondist/index.jsdist/setup.jsdist/hud/hooks.jsdist/hud/autostart.jstemplates/codex-skills/echomem-save/SKILL.md~/.codex/config.toml~/.codex/AGENTS.md~/.codex/skills/echomem-*~/.codex/hooks.json~/.claude/CLAUDE.md~/.claude/settings.json~/Library/LaunchAgents/com.echomem.hud.plist
Network endpoints2
echo-mem-chrome.vercel.appregistry.npmjs.org/%40echomem%2Fmcp/latest
Decision evidence
public snapshotAI called this Suspicious at 90.0% confidence as Benign with medium false-positive risk.
Evidence for warning
- dist/setup.js: init/setup writes MCP entries, global agent guidance, bundled skills, and optional hooks.
- The injected global guidance directs agents to proactively search and save through EchoMem.
- dist/hud/hooks.js installs command hooks into Codex and Claude Code configuration.
- dist/hud/autostart.js can create a macOS LaunchAgent via an explicit CLI command.
Evidence against
- package.json has no preinstall, install, postinstall, or prepare lifecycle hook.
- All agent-control and persistence writes are reached through explicit init/setup or HUD CLI commands.
- No stealth, remote payload execution, credential harvesting, or destructive filesystem chain was found.
- dist/index.js network use is the package's authenticated EchoMem memory API.
Behavioral surface
ChildProcessCryptoEnvironmentVarsFilesystemNetworkShell
HighEntropyStringsMinified
Source & flagged code
6 flaggeddist/forensics.jsView file
19import path from "node:path";
L20: import { execFileSync } from "node:child_process";
L21: import { calculateContextMetrics } from "./context-metrics/calculate.js";
High
dist/hud/autostart.jsView file
3import path from "node:path";
L4: import { spawnSync } from "node:child_process";
L5: // Show-after-restart for the HUD via a macOS LaunchAgent. We ship a CLI, not an .app bundle, so this is
...
L9: export function autostartPlistPath() {
L10: return path.join(os.homedir(), "Library", "LaunchAgents", `${LABEL}.plist`);
L11: }
L12: export function autostartSupported() {
L13: return process.platform === "darwin";
L14: }
...
L20: }
L21: function launchctlUid() {
L22: return typeof process.getuid === "function" ? process.getuid() : 501;
Medium
Install Persistence
Source writes installer persistence such as shell profile or service configuration.
dist/hud/autostart.jsView on unpkg · L3Findings
5 High5 Medium5 Low
HighChild Processdist/forensics.js
HighShell
HighSame File Env Network Executiondist/hud/server.js
HighExternal Ai Agent Control Surface Mutationdist/hud/hooks.js
HighPrevious Version Dangerous Deltadist/setup.js
MediumNetwork
MediumEnvironment Vars
MediumInstall Persistencedist/hud/autostart.js
MediumShips Wasm Moduledist/city/vendor/rive.wasm
MediumStructural Risk Force Deep Review
LowScripts Present
LowFilesystem
LowHigh Entropy Strings
LowUrl Strings
LowNo License
Affected versions and remediation
This report applies to @echomem/mcp@1.4.36.
See all versions of @echomem/mcp for other recorded verdicts.
- Review the evidence and your use of @echomem/mcp@1.4.36 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.