Loading npm security reports…

devcodex@1.17.5: AI-agent control-surface risk | LPM Firewall

registry / devcodex / 1.17.5

devcodex@1.17.5

DevCodex — AI-powered development workflow rules for Copilot, Claude Code, Codex, Gemini CLI, Grok and Cursor

AI Security Review

scanned 1d ago · by lpm-firewall-ai

LPM flags this version as an AI-agent control-surface risk. On npm global installation, the postinstall hook automatically installs DevCodex hooks, instructions, plugins, and local MCP server entries across several user-level AI-agent configurations. This is an unconsented mutation of broad foreign agent control surfaces.

Static reason

High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review

Trigger

npm install -g devcodex

Impact

Subsequent AI-agent sessions can automatically load package-controlled instructions, hooks, and local MCP processes.

Mechanism

postinstall commits host configuration and Node MCP/hook registrations

Policy narrative

The package’s postinstall path executes during a global npm install and applies configuration to all supported host targets by default. Those targets include user-level Codex, Claude, Copilot, Gemini, Grok, and Cursor locations. The generated configuration registers package-controlled hooks and local Node MCP servers, causing future agent activity to invoke package code without an explicit DevCodex command after installation.

Rationale

Although no network exfiltration was found, the package concretely performs unconsented postinstall mutation of broad third-party AI-agent control surfaces. This meets the firewall block policy. Product guard normalized a concrete AI-agent control hijack publish_block to the blockable dangerous-capability shape.

Evidence

package.jsonscripts/postinstall.jsscripts/lib/npm-lifecycle-adapter.jsscripts/lib/global-host-target.jsscripts/lib/global-host-config.js

Source & flagged code

12 flagged · loading source

package.jsonView file

•scripts registry_only=benchmark:execution-chain,benchmark:report-index-wide,benchmark:runtime-indexes,check:control-content,check:skill-intents,check:turn-liveness,cleanup:package-projection,content-root-delete changed=validate

Critical

Manifest Confusion

Tarball package.json differs from the npm registry version manifest for scripts or dependency sets.

package.jsonView on unpkg

•scripts.postinstall = node scripts/postinstall.js

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg

•scripts.postinstall = node scripts/postinstall.js

Medium

Ambiguous Install Lifecycle Script

Install-time lifecycle script is not statically allowlisted and needs review.

package.jsonView on unpkg

index.jsView file

8L9: const fs = require('fs') L10: const os = require('os')

Medium

Dynamic Require

Package source references dynamic require/import behavior.

index.jsView on unpkg · L8

scripts/lib/global-host-config.jsView file

1Install-time code directly mutates a foreign AI-agent control surface: L227: addInstructionRoot(operations, target.host, packageRoot, path.join(runtime, 'instructions.full.md'), fsImpl) L228: addSourceFile(operations, target.host, path.join(packageRoot, 'host-projections', 'AGENTS.md'), path.join(runtime, 'AGENTS.md'), fsImpl) L229: addSourceTree(operations, target.host, path.join(packageRoot, 'hooks', '_runtime'), path.join(runtime, 'hooks', '_runtime'), fsImpl) ... L333: * Map global host id → memory/profile agent identity (VALID_AGENTS). L334: * Claude host root is `.claude` but agent id is `claude-code`. L335: */ ... L343: L344: function buildMcpServers(runtimeRoot, options = {}) { L345: const agent = hostToRuntimeAgent(options.agent || options.host) ... L365: L366: function buildCop

Critical

Ai Agent Control Hijack

Source creates an unconsented AI-agent control surface through install-time mutation or a default unauthenticated remote skill channel.

scripts/lib/global-host-config.jsView on unpkg · L1

scripts/lib/host-adapter-scope.jsView file

•Manifest-trigger-reachable source links an external AI-agent control path to a behavior-bearing write operation. anned-install') : (after.repoid && sourcedigest === installeddigest ? 'verified' : 'mismatch') if (!dryrun && status !== 'verified') { const error = new error('grok_plugin_install_mismatch: installed plugin bytes differ from workspace owner') error.code = 'grok_plugin_install_mismatch' throw error } return { schemaversion: 'grokplugininstallationreceiptv1', status, pluginpath: portable(source), repoid: after.repoid, installedpath: installedpath ? portable(installedpath) : null, sourcedigest, installeddigest, grokversion: string(probe.stdout || probe.stderr || '').trim(), refreshmode, dryrun } } function timestampsu

High

Trigger Reachable External Ai Agent Control Surface Mutation

Manifest-trigger-reachable source writes behavior-bearing configuration into a user or project AI-agent control surface.

scripts/lib/host-adapter-scope.jsView on unpkg

hooks/_runtime/stdio-bounds.cjsView file

•stage = ast_semantic_analysis; reason = ast_alias_growth_limit_exceeded; limitedFiles = 2

High

Semantic Analysis Limited

A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.

hooks/_runtime/stdio-bounds.cjsView on unpkg

scripts/lib/global-host-runtime-verifier.jsView file

•matchType = normalized_sha256 matchedPackage = devcodex@1.17.3 matchedPath = scripts/lib/global-host-runtime-verifier.js matchedIdentity = npm:ZGV2Y29kZXg:1.17.3 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

scripts/lib/global-host-runtime-verifier.jsView on unpkg

scripts/probe-skill-route-s15-host.jsView file

•matchType = normalized_sha256 matchedPackage = devcodex@1.17.3 matchedPath = scripts/probe-sk[redacted] matchedIdentity = npm:ZGV2Y29kZXg:1.17.3 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

scripts/probe-skill-route-s15-host.jsView on unpkg

scripts/test-runtime-rolling-upgrade.jsView file

•matchType = normalized_sha256 matchedPackage = devcodex@1.17.3 matchedPath = scripts/test-runtime-rolling-upgrade.js matchedIdentity = npm:ZGV2Y29kZXg:1.17.3 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

scripts/test-runtime-rolling-upgrade.jsView on unpkg

grok/plugins/devcodex-workspace/hooks/devcodex-workspace.cjsView file

•matchType = normalized_sha256 matchedPackage = devcodex@1.17.3 matchedPath = grok/plugins/devcodex-workspace/hooks/devcodex-workspace.cjs matchedIdentity = npm:ZGV2Y29kZXg:1.17.3 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

grok/plugins/devcodex-workspace/hooks/devcodex-workspace.cjsView on unpkg

scripts/lib/node-runtime-readiness.jsView file

•matchType = normalized_sha256 matchedPackage = devcodex@1.17.3 matchedPath = scripts/lib/node-runtime-readiness.js matchedIdentity = npm:ZGV2Y29kZXg:1.17.3 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

scripts/lib/node-runtime-readiness.jsView on unpkg