registry / mtech-ai-local-runner / 0.1.55
mtech-ai-local-runner@0.1.55
⚠ Under reviewThe guide integration (Work Item d38d60f9) adds two hooks to Claude Code and Codex. Install it only on request:
AI Security Review
scanned 10d ago · by lpm-firewall-aiThe source shows pinned native helper execution, but both bundled helpers remain uninterpretable in the available analyses. That unresolved behavior prevents a clean verdict, while the inspected source does not independently establish malicious behavior.
Static reason
No blocking static signals were detected.
Trigger
A Codex observation or policy inventory request reaches the helper execution path.
Impact
The effect of the native helpers cannot be determined from the available evidence.
Mechanism
The package verifies a pinned helper and invokes it as a subprocess; the available artifact analyses could not determine the helpers' behavior.
AI rationale
The source shows pinned native helper execution, but both bundled helpers remain uninterpretable in the available analyses. That unresolved behavior prevents a clean verdict, while the inspected source does not independently establish malicious behavior.
Evidence
native/bin/codex-observe-darwin-arm64native/bin/policy-observer-darwin-arm64.lpm-inspection-hrOWni/0-681fb12e0b83.txt.lpm-inspection-hrOWni/1-2a07812cd477.txtdist/src/safety/policy-inventory-client.jsdist/src/safety/codex-manual-registered.js
Decision evidence
public snapshotAI called this Manual Review at 92.0% confidence as Unknown with medium false-positive risk.
AI review observations
- The Codex integration code uses atomic file writes, and its caller invokes installation only for the explicit install action.
- The package includes two native executables whose behavior remains unresolved.
- The policy-inventory client executes a binary, while registered source pins a privileged helper path to a hash.
- The client checks the executable against its pinned hash before use.
Evidence against
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsFilesystemNetworkWebSocket
HighEntropyStringsUrlStrings
NoLicense
Source & flagged code
5 flaggedREADME.mdView file
•carries validated ids only; the AI reads the ask itself with the Gateway show tools, which mark its text as untrusted.
High
Ai Reviewer Manipulation
Package text addresses the security reviewer or scanner and tries to influence the review outcome.
README.mdView on unpkgdist/src/adapters/codex/pre-model-gate.jsView file
29reasonCodes.push(code); };
L30: require(actual.model === expected.model && actual.provider === expected.provider && actual.cwd === expected.cwd, 'RUNTIME_IDENTITY_MISMATCH');
L31: require(actual.configFingerprint === expected.configFingerprint, 'CONFIG_FINGERPRINT_MISMATCH');
Medium
Dynamic Require
Package source references dynamic require/import behavior.
dist/src/adapters/codex/pre-model-gate.jsView on unpkg · L29dist/src/publication/raw-git-objects.jsView file
Findings
2 High5 Medium6 Low
HighAi Reviewer ManipulationREADME.md
HighSemantic Analysis Limiteddist/src/asks/doctor.js#virtual:normalized:round1
MediumDynamic Requiredist/src/adapters/codex/pre-model-gate.js
MediumNetwork
MediumEnvironment Vars
MediumShips Native Binarynative/bin/codex-observe-darwin-arm64
MediumStructural Risk Force Deep Review
LowScripts Present
LowWeak Cryptodist/src/publication/raw-git-objects.js
LowFilesystem
LowHigh Entropy Strings
LowUrl Strings
LowNo License
Affected versions and remediation
This report applies to mtech-ai-local-runner@0.1.55.
See all versions of mtech-ai-local-runner for other recorded verdicts.
- Review the evidence and your use of mtech-ai-local-runner@0.1.55 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.