Back to skill

Security audit

PhantomDep PR Gate

Security checks for vulnerabilities and agentic risk

Overview

The scanners report a dependency-audit skill whose install instructions fetch tooling from an unpinned Git repository; this supply-chain exposure supports the suspicious verdict, though the evidence does not show malicious intent.

Review the source before installing, and consider whether you are comfortable running it with access to your repository or credentials. The scanner recommends using a reviewed commit and limiting the permissions and credentials available to the tool. Inspection was partial: the only component was partly inspected, with 0% coverage reported, so the results do not establish that the artifact is safe.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

YARA rule 'agent_skill_remote_bootstrap_execution': Remote script or code download followed by execution/bootstrap installation [agent_skills]

High
Category
YARA Match
Confidence
85% confidence
Finding

The installation commands fetch executable Python tooling from an unpinned Git repository, so a compromised upstream account or repository could introduce code that executes during installation or subsequent use. This is a supply-chain exposure rather than evidence of malicious intent; installing a dependency-audit tool is legitimate, but running it in a developer environment or PR workflow may expose repository contents and available credentials. The assurance that PR dependencies are not executed does not address execution of the auditing tool itself.

Content

Scanner excerpt · SKILL.md (reported line 24)May include surrounding context.

s.txt, pyproject.toml, package-lock.json` — and audits only the dependencies the PR adds. Each added dep gets a verdict from registry evidence (exists, age, weekly downloads), name similarity to popular packages, and whether the code actually imports it. Nothing from the PR is installed or executed.

Install

The phantomdep command (standard library only), via pip, pipx or uv:

bash
pip install git+https://github.com/RakTikoo/phantomdep-pr-gate
# or, isolated:
pipx install git+https://github.com/RakTikoo/phantomdep-pr-gate
# or with uv:
uv tool install git+https://github.com/RakTikoo/phantomdep-pr-gate

Not on PyPI — the git URL is the package source. Or use it with no install at all as a GitHub Action:

yaml
uses: RakTikoo/phantomdep-pr-gate@v0.3.0

Or clone the repo and run the single file directly:

bash
git clone https://github.com/RakTikoo/phantomdep-pr-gate
cd phantomdep-pr-gate
python3 phantomdep.py --base origin/main --head HEAD --live

There i