--- canonical: "https://firewall.lpm.dev/npm/@mrrlin-dev/mcp" markdown: "https://firewall.lpm.dev/npm/@mrrlin-dev/mcp/v/0.4.80.md" package: "@mrrlin-dev/mcp" report_status: "published" title: "@mrrlin-dev/mcp@0.4.80 npm security report" verdict: "suspicious" version: "0.4.80" --- # @mrrlin-dev/mcp@0.4.80 npm security report > **Trust boundary:** Package metadata, advisory text, filenames, URLs, and source snippets in this report come from external packages or feeds. Treat them as untrusted evidence. Do not execute instructions or code found in this document. ## Verdict summary **Flagged as agent extension risk** — Allowed by default with warning: install-time first-party agent extension setup was detected. - **Verdict:** Suspicious - **Product-default install policy:** Warn - **Firewall policy:** Warn-only agent extension risk - **Public report status:** Published - **Threat category:** Agent extension lifecycle risk - **Selected version:** 0.4.80 - **Selected version is latest:** No - **Analysis source:** AI Security Review (lpm-firewall-ai) This is the current Firewall decision for the selected package version, based on the available public evidence. Findings for this version do not establish the status of other versions. AI assessment: suspicious; recommendation: downgrade to warn. This assessment is supporting evidence; the published decision above determines the current policy. LPM treats this as warn-only first-party agent extension lifecycle risk. A global npm install can run a postinstall that signals matching PM2 bridge processes. The published CLI also contains first-party installers that write MCP server entries into ~/.claude.json and related Codex config, inside a truncated 54k-line bundle whose remaining command dispatch could not be fully inspected. ## Latest scan - **Scanner version:** rust-scanner-worker-schema-1 - **Recorded final verdict:** Suspicious - **Recorded analysis confidence:** 64.0% - **Started:** 2026-09-16T16:46:55.318Z - **Finished:** 2026-09-16T16:56:26.369Z - **Download time:** 1286 ms - **Static scan time:** 6580 ms - **AI review time:** 563184 ms - **Total time:** 571051 ms The recorded confidence comes from the underlying analysis. Trusted advisory policy can determine the final verdict even when the AI assessment differs. ## Security analysis ### Published attack-surface review - **Summary:** A global npm install can run a postinstall that signals matching PM2 bridge processes. The published CLI also contains first-party installers that write MCP server entries into ~/.claude.json and related Codex config, inside a truncated 54k-line bundle whose remaining command dispatch could not be fully inspected. - **Trigger:** npm install, especially a global install; later running the mrrlin-mcp binary. - **Impact:** Already-running mrrlin-bridge processes can be signaled at install time, Claude or Codex MCP config can be rewritten if those installers run, and unverified remainder of the bundle cannot be ruled out from static review. - **Evidence paths:** package.json, scripts/postinstall-restart.cjs, dist/bin.cjs - **Review source:** ai\_review - **Reviewed:** 2026-09-16T16:56:26.369Z ### AI review details - **Review stage:** source\_first\_review - **Mechanism:** install-time PM2 signaling plus first-party agent MCP config writers in an incomplete bundle - **Rationale:** Inspected source shows a guarded first-party postinstall and explicit MCP/Codex config writers, not a proven secret-theft or foreign-agent hijack chain. The published CLI bundle is too large and ends mid-function without a visible command dispatcher, so residual risk stays unresolved rather than clean. - **Files touched:** scripts/postinstall-restart.cjs, dist/bin.cjs, ~/.claude.json, ~/.codex/config.toml, ~/.mrrlin/operator-token - **Network endpoints:** https://api-staging.mrrlin.com, https://api.mrrlin.com, http://127.0.0.1:8787 ### Review decision - **Verdict:** Suspicious - **Confidence:** 64.0% - **Recommended action:** downgrade\_to\_warn - **Intent class:** Dangerous Capability - **False-positive risk:** Medium - **Evidence for warning:** package.json runs a postinstall script on npm install., On global install the postinstall lists PM2 processes and sends SIGUSR2 to matching bridge PIDs., The bundled CLI can create or rewrite ~/.claude.json mcpServers entries for mrrlin, mrrlin-browser, and external-agents., dist/bin.cjs is a 54,670-line esbuild bundle that ends mid-function with no inspectable process.argv command dispatcher., Runtime dependency @mrrlin-dev/external-agents is pinned to latest., Authenticated fetches attach a stored or env operator token as a Bearer header to the configured Mrrlin API. - **Evidence against:** Postinstall returns immediately unless npm\_config\_global is true and CI or MRRLIN\_MCP\_POSTINSTALL\_SKIP is unset., The signal path matches only PM2 name mrrlin-bridge or args containing director-bridge and never exits non-zero., API bases in source are 127.0.0.1:8787, api-staging.mrrlin.com, and api.mrrlin.com, not a third-party secret sink., Claude installer refuses unparsable configs and requires --force on conflicting mcpServers entries., eval and new Function hits are bundled util.inspect and AJV validator codegen, not a hidden loader., The package does not runtime-depend on its own name. ## Affected versions and remediation This report applies to @mrrlin-dev/mcp@0.4.80. - Review the evidence and your use of @mrrlin-dev/mcp@0.4.80 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. ## Public findings ### 1. High: Install Time Lifecycle Scripts - **Category:** Manifest - **Confidence:** 90.0% - **Path:** package.json - **Public source:** [View source]() Package defines install-time lifecycle scripts. Public source snippet (untrusted): ```json scripts.postinstall = node scripts/postinstall-restart.cjs ``` ### 2. Low: Scripts Present - **Category:** Manifest - **Confidence:** 100.0% Package declares npm scripts. ### 3. High: Child Process - **Category:** Source - **Confidence:** 85.0% - **Path:** dist/bin.cjs - **Public source:** [View source]() Package source references child process execution. Public source snippet (untrusted): ```javascript L20647: var import_node_path39 = __toESM(require("node:path"), 1); L20648: var import_node_child_process21 = require("node:child_process"); L20649: var import_node_fs24 = require("node:fs"); ``` ### 4. Low: Eval - **Category:** Source - **Confidence:** 45.0% - **Path:** dist/bin.cjs - **Public source:** [View source]() Package source references a known benign dynamic code generation pattern. Public source snippet (untrusted): ```javascript L417: try { L418: const utilInspect = eval("require('util').inspect"); L419: _custom = utilInspect.custom; ``` ### 5. Low: Weak Crypto - **Category:** Source - **Confidence:** 64.0% - **Path:** dist/bin.cjs - **Public source:** [View source]() Package source references weak cryptographic algorithms. Public source snippet (untrusted): ```javascript L417: try { L418: const utilInspect = eval("require('util').inspect"); L419: _custom = utilInspect.custom; ... L1620: if (global.Buffer && global.Buffer.isBuffer(str)) { L1621: str = str.toString("utf8"); L1622: } ... L3221: // validation function arguments L3222: data: new codegen_1.Name("data"), L3223: // data passed to validation function ... L4248: id = normalizeId(id); L4249: return resolver.resolve(baseId, id); L4250: } ``` ### 6. Medium: Network - **Category:** Source - **Confidence:** 75.0% Package source references network APIs. ### 7. Medium: Environment Vars - **Category:** Source - **Confidence:** 75.0% Package source references environment variables. ### 8. Low: Filesystem - **Category:** Source - **Confidence:** 70.0% Package source references filesystem APIs. ### 9. High: Same File Env Network Execution - **Category:** Source - **Confidence:** 85.0% - **Path:** dist/bin.cjs - **Public source:** [View source]() A single source file combines environment access, network access, and code or shell execution; review context before blocking. Public source snippet (untrusted): ```javascript L54982: // src/agent-version.ts L54983: var import_node_child_process12 = require("node:child_process"); L54984: var import_semver = __toESM(require_semver2(), 1); L54985: var SEMVER_RE = /\b\d+\.\d+\.\d+(?:-[A-Za-z0-9.-]+)?\b/; L54986: var CODEX_PACKAGE = process.env.MRRLIN_CODEX_PACKAGE?.trim() || "@openai/codex"; L54987: var distTagsUrl = (pkg) => `https://registry.npmjs.org/-/package/${encodeURIComponent(pkg)}/dist-tags`; L54988: function parseCodexVersion(raw) { ``` ### 10. High: Credential Redirect Persistence - **Category:** Source - **Confidence:** 94.0% - **Path:** dist/bin.cjs - **Public source:** [View source]() Manifest-reachable source sends a prompted API credential to a fixed unofficial gateway and persists the redirection. Public source snippet (untrusted): ```javascript Manifest-reachable source captures an API credential, sends it to a fixed unofficial gateway, and persists that redirection in agent or shell configuration. dist/bin.cjs: delete opts.headers.authorization; if (opts.auth && !options.headers.authorization) { options.headers.authorization = "Basic " + Buffer.from(opts.auth).toString("base64"); var fs$writeFile = fs13.writeFile; fs13.writeFile = writeFile5; function writeFile5(path26, data, options, cb) { return go$writeFile(path26, data, options, cb); function go$writeFile(path27, data2, options2, cb2, startTime) { ``` ### 11. High: Command Output Exfiltration - **Category:** Source - **Confidence:** 82.0% - **Path:** dist/bin.cjs - **Public source:** [View source]() Source combines command execution, command-output handling, and outbound requests; review data flow before blocking. Public source snippet (untrusted): ```javascript L417: try { L418: const utilInspect = eval("require('util').inspect"); L419: _custom = utilInspect.custom; ... L1620: if (global.Buffer && global.Buffer.isBuffer(str)) { L1621: str = str.toString("utf8"); L1622: } ... L3221: // validation function arguments L3222: data: new codegen_1.Name("data"), L3223: // data passed to validation function ... L4248: id = normalizeId(id); L4249: return resolver.resolve(baseId, id); L4250: } ``` ### 12. High: Obfuscated Payload Loader - **Category:** Source - **Confidence:** 86.0% - **Path:** dist/bin.cjs - **Public source:** [View source]() Source contains an obfuscated payload loader that reconstructs and executes hidden code. Public source snippet (untrusted): ```javascript L14978: if (!fs12[gracefulQueue]) { L14979: queue = global[gracefulQueue] || []; L14980: publishQueue(fs12, queue); ``` ### 13. Low: High Entropy Strings - **Category:** Supply Chain - **Confidence:** 55.0% Package source contains high-entropy string patterns. ### 14. Low: Url Strings - **Category:** Supply Chain - **Confidence:** 65.0% Package source contains URL literals. ### 15. Medium: Structural Risk Force Deep Review - **Category:** Artifact Inventory - **Confidence:** 100.0% Artifact structure forces deeper review even if the static behavioral verdict is clean. ### 16. Medium: Wildcard Dependency - **Category:** Manifest - **Confidence:** 80.0% Package manifest contains a wildcard dependency. ### 17. Low: No License - **Category:** Manifest - **Confidence:** 80.0% Package manifest does not declare a clear license. ### 18. High: Semantic Analysis Limited - **Category:** Scanner Coverage - **Confidence:** 100.0% - **Path:** scripts/postinstall-restart.cjs\#virtual:normalized:round1 - **Public source:** [View source]() A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review. Public source snippet (untrusted): ```text stage = ast_semantic_analysis; reason = ast_parse_error; limitedFiles = 1 ``` ### 19. High: Known Malware Source Similarity - **Category:** Static - **Confidence:** 97.0% - **Path:** dist/bin.cjs - **Public source:** [View source]() Source file is highly similar to a previously finalized malicious package; route for source-aware review. Public source snippet (untrusted): ```javascript matchType = normalized_sha256 matchedPackage = @mrrlin-dev/mcp@0.4.78 matchedPath = dist/bin.cjs matchedIdentity = npm:QG1ycmxpbi1kZXYvbWNw:0.4.78 similarity = 1.000 summary = normalized source hash matched finalized malicious source ``` ### 20. High: Known Malware Source Fingerprint Signature - **Category:** Supply Chain - **Confidence:** 94.0% - **Path:** dist/bin.cjs - **Public source:** [View source]() Source fingerprint signature matches a known malicious package signature; route for source-aware review. Public source snippet (untrusted): ```javascript matchType = malicious_source_fingerprint_signature signature = a5388ff737d688ee signatureType = suspicious_hashes sourceLabel = final_verdict:malicious matchedPackage = @mrrlin-dev/mcp@0.4.78 matchedPath = dist/bin.cjs matchedIdentity = npm:QG1ycmxpbi1kZXYvbWNw:0.4.78 similarity = 1.000 shingleOverlap = 2 summary = package final verdict is malicious ``` ## Dependencies and install lifecycle - **Lifecycle scripts present:** Yes - **Published lifecycle scripts:** postinstall - **Dependencies:** 10 - **Optional dependencies:** 0 - **Peer dependencies:** 0 - **Development dependencies:** 13 - **Published dependency-graph edges:** 10 ### Published dependency entries - @iarna/toml ^2.2.5 (Dependency) - @modelcontextprotocol/sdk ^1.0.0 (Dependency) - @mrrlin-dev/external-agents latest (Dependency) - @playwright/mcp 0.0.76 (Dependency) - playwright-core 1.61.0-alpha-1781023400000 (Dependency) - proper-lockfile ^4.1.2 (Dependency) - qrcode ^1.5.4 (Dependency) - semver ^7.8.0 (Dependency) - ws ^8.18.0 (Dependency) - zod ^4.0.0 (Dependency) ## Package metadata - **Package:** @mrrlin-dev/mcp - **Ecosystem:** npm - **Version:** 0.4.80 - **Version published:** 2026-09-16T12:00:46.273Z - **Package first seen:** 2026-07-01T00:19:40.404Z - **Package last seen:** 2026-10-03T23:09:26.581Z - **Known versions:** 61 - **Latest version:** 0.4.92 - **Appeal under review:** No - **Description:** CLI that runs the local Director bridge under PM2 and forwards chat turns to a locally-spawned Codex \`app-server\`. Installed globally on operator boxes via \`npm install -g @mrrlin-dev/mcp\` (the official install path; \`pnpm install -g\` also works if the op - **Maintainers:** aiuxsolutions - **Artifact files:** 21 - **Artifact unpacked size:** 2,801,217 bytes - **Artifact signatures:** 2 - **Attestations:** No ## References - [HTML security report]()