Security audit

logsentry

Security checks across malware telemetry and agentic risk

Overview

LogSentry is mostly a coherent local scanner, but it needs Review because a crafted license key can trigger local code execution and optional git hooks persist in repositories.

Install only if you trust the publisher and treat license keys as sensitive executable-adjacent input. Prefer environment or locked-down config over CLI flags, avoid untrusted license keys, and use hooks only as an explicit opt-in because they modify your repository and run on future commits or pushes.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (9)

Tp4

High

Category: MCP Tool Poisoning
Confidence: 84% confidence
Finding: The declared description presents the skill as a passive logging analyzer, but the documented behavior extends into local configuration access, license processing, and especially git-hook installation/modification via lefthook. This mismatch can cause users or orchestrators to invoke a skill expecting read-only analysis while it performs repository-changing actions, increasing the risk of unintended persistence or workflow modification.

Description-Behavior Mismatch

Medium

Confidence: 93% confidence
Finding: The dispatcher includes hook installation and repository configuration editing, which exceeds the stated purpose of a logging analyzer and gives the skill write access to project configuration. While not inherently malicious, this expands the trust boundary and can surprise users by persisting behavior into their repositories.

Context-Inappropriate Capability

High

Confidence: 95% confidence
Finding: This code modifies lefthook.yml in the user's repository and installs git hooks, creating persistent execution on future commits. For a logging analyzer, that is a significant capability escalation because it alters developer workflow and can be abused to run arbitrary code in trusted repo contexts if the skill or referenced path is later changed.

Context-Inappropriate Capability

Medium

Confidence: 82% confidence
Finding: The script reads a license/API key from an environment variable and a local config file even though the advertised skill purpose is logging analysis, which expands privilege and secret-access scope beyond what users may expect. In an agent setting, unnecessary secret access is dangerous because any component that reads sensitive material can later expose it via logs, errors, subprocess arguments, or future code changes, increasing the blast radius of compromise or misuse.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The README explicitly demonstrates supplying a license key via the `--license-key` command-line flag. Secrets passed as CLI arguments can be exposed through shell history, process listings, CI job diagnostics, or audit tooling, so this documentation encourages an unsafe secret-handling pattern even if the tool itself is local.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The configuration example shows storing a license key in plaintext in `~/.openclaw/openclaw.json` without any caution about credential sensitivity or file permissions. This can lead users to persist secrets in world-readable or broadly accessible files, increasing risk of credential disclosure on shared systems, backups, or source-controlled configs.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The CI example injects a secret into a shell command without any guidance on preventing disclosure through verbose shell output, debug logging, or workflow misconfiguration. While CI secret stores help, README examples that interpolate secrets into commands can still normalize practices that leak credentials in logs or troubleshooting output.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: Passing a license key via command-line argument is risky because command-line parameters are commonly exposed through shell history, process listings, CI logs, and telemetry. Exporting that value to the environment increases propagation without warning the user about the exposure risk.

Missing User Warnings

Low

Confidence: 78% confidence
Finding: The script accesses secrets from the environment and user config without clear runtime disclosure or consent, which is a transparency and least-surprise problem in a skill context. This becomes more concerning because the skill is framed as an observability analyzer, so users may not anticipate credential access, making accidental overtrust and secret exposure more likely.

VirusTotal

61/61 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.