--- canonical: "https://firewall.lpm.dev/npm/@theronap/cortex-mcp/v/0.9.122" markdown: "https://firewall.lpm.dev/npm/@theronap/cortex-mcp/v/0.9.122.md" package: "@theronap/cortex-mcp" report_status: "published" title: "@theronap/cortex-mcp@0.9.122 npm security report" verdict: "suspicious" version: "0.9.122" --- # @theronap/cortex-mcp@0.9.122 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:** 0.9.122 - **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. This package creates durable AI-agent hooks and can transmit session-derived content after explicit user setup, so it warrants a warning. No lifecycle script, hidden install execution, self-dependency chain, or concrete malicious behavior was found. ## Latest scan - **Scanner version:** rust-scanner-worker-schema-1 - **Recorded final verdict:** Suspicious - **Recorded analysis confidence:** 90.0% - **Started:** 2026-08-30T00:43:11.953Z - **Finished:** 2026-08-30T00:44:33.540Z - **Download time:** 502 ms - **Static scan time:** 1056 ms - **AI review time:** 80028 ms - **Total time:** 81587 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:** After an explicit setup or install command, the package persists MCP server entries and hooks in AI-agent configuration. Those hooks can collect session content, contact Cortex, and refresh remotely served skills on later agent events. - **Trigger:** The user runs cortex-mcp setup or install; configured Stop and SessionStart hooks then run during later agent sessions. - **Impact:** Session-derived content can be sent to the configured Cortex service, and organization-provided skill text can be installed into agent skill directories. No automatic npm install-time trigger was found. - **Evidence paths:** package.json, bin/cortex-mcp.mjs, lib/setup.mjs, lib/editors/claude.mjs, lib/capture.mjs, lib/skills.mjs, lib/editors/antigravity.mjs - **Review source:** ai\_review - **Reviewed:** 2026-08-30T00:44:33.540Z ### AI review details - **Review stage:** source\_first\_review - **Mechanism:** Explicit agent configuration, hook-based session capture, and remote skill synchronization. - **Rationale:** This package creates durable AI-agent hooks and can transmit session-derived content after explicit user setup, so it warrants a warning. No lifecycle script, hidden install execution, self-dependency chain, or concrete malicious behavior was found. - **Files touched:** ~/.claude.json, ~/.claude/settings.json, ~/.codex/config.toml, ~/.codex/hooks.json, ~/.cortex/capture.log, ~/.claude/skills, ~/.codex/skills - **Network endpoints:** https://cortex-console.vercel.app/api/ingest, https://cortex-console.vercel.app/api/skills ### Review decision - **Verdict:** Suspicious - **Confidence:** 90.0% - **Recommended action:** downgrade\_to\_warn - **Intent class:** Dangerous Capability - **False-positive risk:** Low - **Evidence for warning:** Explicit setup writes MCP credentials and recurring hooks into AI-agent configuration., The installed Stop hook reads a session transcript and posts a derived digest or fallback transcript tail to the Cortex API., Session-start repair fetches organization-provided skills and writes them into installed agent skill directories., The Claude setup adds an allowlist for the package's MCP tools, including authoring actions., The Antigravity adapter can create a launchd watcher artifact, though it does not load it. - **Evidence against:** package.json has no preinstall, install, or postinstall script., All configuration changes require an explicit CLI setup or install command; no package import performs them., Network use is consistently directed to the package's configured Cortex service and authenticated with its token., The capture path redacts credential-shaped text before transmission. ## Affected versions and remediation This report applies to @theronap/cortex-mcp@0.9.122. - Review the evidence and your use of @theronap/cortex-mcp@0.9.122 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. Low: Scripts Present - **Category:** Manifest - **Confidence:** 100.0% Package declares npm scripts. ### 2. Low: Weak Crypto - **Category:** Source - **Confidence:** 64.0% - **Path:** lib/capture.mjs - **Public source:** [View source]() Package source references weak cryptographic algorithms. Public source snippet (untrusted): ```javascript L1: import { readFileSync, accessSync, constants, openSync, mkdirSync } from 'fs' L2: import { spawn, execFileSync } from 'child_process' L3: import { homedir } from 'os' ... L17: if (!res.ok) return null L18: const j = await res.json().catch(() => ({})) L19: return Array.isArray(j.types) ? j.types : null ... L31: try { L32: if (resolve(String(cwd)) === homedir()) return 'general' L33: } catch { /* unresolvable path — fall through to basename */ } ... L41: // L42: // Handles the four remote forms git emits: scp-like ssh (git@github.com:o/n.git), ssh://, https://, L43: // git://. The HOST CHECK IS LOAD-BEARING, not cosmetic: this string becomes a clearance key, and ``` ### 3. Medium: Network - **Category:** Source - **Confidence:** 75.0% Package source references network APIs. ### 4. Medium: Environment Vars - **Category:** Source - **Confidence:** 75.0% Package source references environment variables. ### 5. Medium: Install Persistence - **Category:** Source - **Confidence:** 75.0% - **Path:** lib/editors/antigravity.mjs - **Public source:** [View source]() Source writes installer persistence such as shell profile or service configuration. Public source snippet (untrusted): ```javascript L4: // WatchPaths the trajectory store and runs a one-shot sync. wire() generates the two artifacts L5: // (the sync.sh + the launchd plist) deterministically. It does NOT run `launchctl load` here L6: // (an unverifiable system mutation) and it flags the PORTABILITY BLOCKER below. ... L21: /** The launchd plist, home-substituted. Deterministic — golden-testable byte-for-byte. */ L22: export function renderAntigravityPlist({ home = homedir() } = {}) { L23: const sh = join(home, '.cortex', 'antigravity-sync.sh') ... L33: L34: /bin/bash L35: ${sh} ``` ### 6. Low: Filesystem - **Category:** Source - **Confidence:** 70.0% Package source references filesystem APIs. ### 7. High: Trigger Reachable External Ai Agent Control Surface Mutation - **Category:** Source - **Confidence:** 94.0% - **Path:** lib/uninstall.mjs - **Public source:** [View source]() Manifest-trigger-reachable source writes behavior-bearing configuration into a user or project AI-agent control surface. Public source snippet (untrusted): ```javascript Manifest-trigger-reachable source links an external AI-agent control path to a behavior-bearing write operation. import { readfilesync, writefilesync, existssync, copyfilesync, rmsync } from 'fs' import { homedir } from 'os' import { join } from 'path' import { execfilesync } from 'child_process' import { ismanagedcommand } from './managed.mjs' ``` ### 8. High: Trigger Reachable Persistence - **Category:** Source - **Confidence:** 94.0% - **Path:** lib/editors/antigravity.mjs - **Public source:** [View source]() A manifest entrypoint or package-local install chain reaches persistence behavior. Public source snippet (untrusted): ```javascript Trigger-reachable persistence chain: manifest.bin -> bin/cortex-mcp.mjs -> lib/install.mjs -> lib/editors/index.mjs -> lib/editors/antigravity.mjs L4: // WatchPaths the trajectory store and runs a one-shot sync. wire() generates the two artifacts L5: // (the sync.sh + the launchd plist) deterministically. It does NOT run `launchctl load` here L6: // (an unverifiable system mutation) and it flags the PORTABILITY BLOCKER below. ... L21: /** The launchd plist, home-substituted. Deterministic — golden-testable byte-for-byte. */ L22: export function renderAntigravityPlist({ home = homedir() } = {}) { L23: const sh = join(home, '.cortex', 'antigravity-sync.sh') ... L33: L34: /bin/bash L35: ${sh} ``` ### 9. Low: High Entropy Strings - **Category:** Supply Chain - **Confidence:** 55.0% Package source contains high-entropy string patterns. ### 10. Low: Url Strings - **Category:** Supply Chain - **Confidence:** 65.0% Package source contains URL literals. ### 11. Medium: Ai Review Evidence - **Category:** Ai Review - **Confidence:** 90.0% - **Path:** lib/setup.mjs - **Public source:** [View source]() Explicit setup writes MCP credentials and recurring hooks into AI-agent configuration. Public source snippet (untrusted): ```javascript // ── 1. MCP server in ~/.claude.json ────────────────────────────────────── try { let cfg try { cfg = readJson(claudeJson) } catch (e) { process.stderr.write(`\n✗ ${claudeJson} is not valid JSON — fix or remove it, then re-run.\n`) process.exit(1) } const bak = backup(claudeJson) cfg = mergeClaudeMcp(cfg, spec, token) ensureDir(claudeJson) writeFileSync(claudeJson, JSON.stringify(cfg, null, 2)) ``` ### 12. Medium: Ai Review Evidence - **Category:** Ai Review - **Confidence:** 90.0% - **Path:** lib/capture.mjs - **Public source:** [View source]() The installed Stop hook reads a session transcript and posts a derived digest or fallback transcript tail to the Cortex API. Public source snippet (untrusted): ```javascript const repo = projectFrom(hook.cwd) // Redact credential-shaped strings BEFORE anything leaves the machine — this `transcript` // feeds local edge/typed extraction AND the fallback POST to the org. A secret in a // transcript (pasted key, a tool that read a config file, a token inlined in a hook cmd) // must never be transmitted or stored. See redact.mjs. const transcript = hook.transcript_path ? redactSecrets(transcriptTail(hook.transcript_path)) : '' ``` ## Dependencies and install lifecycle - **Lifecycle scripts present:** No - **Dependencies:** 2 - **Optional dependencies:** 0 - **Peer dependencies:** 0 - **Development dependencies:** 0 - **Published dependency-graph edges:** 2 ### Published dependency entries - @modelcontextprotocol/sdk ^1.29.0 (Dependency) - zod ^3.23.8 (Dependency) ## Package metadata - **Package:** @theronap/cortex-mcp - **Ecosystem:** npm - **Version:** 0.9.122 - **License:** MIT - **Version published:** 2026-08-29T23:04:53.832Z - **Package first seen:** 2026-06-30T23:17:22.311Z - **Package last seen:** 2026-10-30T02:04:41.998Z - **Known versions:** 144 - **Latest version:** 0.9.282 - **Appeal under review:** No - **Description:** Connect your AI assistant to Cortex — your org's projects, activity, gaps, and directives, scoped to you. - **Deprecated:** Renamed to @theronap/agnoclast-mcp — use that. Existing installs switch themselves to the new name on next start. - **Maintainers:** theronap - **Keywords:** mcp, cortex, claude, ai, org-intelligence - **Runtime engines:** node: \>=18 - **Artifact files:** 45 - **Artifact unpacked size:** 561,176 bytes - **Artifact signatures:** 1 - **Attestations:** No ## References - [HTML security report]()