Back to skill

Security audit

Single-cell Pipeline

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed bioinformatics template generator, with notable dependency and overwrite hygiene issues but no evidence of hidden, destructive, or deceptive behavior.

Install and run this only in an isolated project environment, review the generated output path before execution, and prefer pinned/locked dependencies. Pay particular attention to the root requirements entry scvi versus the documented scvi-tools before running pip install -r requirements.txt.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (1)

T08 · Insecure Dependencies

Warning
Location
requirements.txt:1
Finding

Unpinned and inconsistent third-party dependencies create supply-chain risk

Content
View full analysis

Vulnerability Details

File Location: requirements.txt:1-9; references/requirements.txt:1-23; SKILL.md:220-227, 294; scripts/main.py:634
Vulnerability Type: Unpinned dependencies, mutable source dependencies, and package-name inconsistency
Risk Level: Medium

Vulnerable Code

requirements.txt:1-9:

text
bbknn
harmonypy
matplotlib
numpy
pandas
scanorama
scanpy
scvi
seaborn

references/requirements.txt:1-23:

text
scanpy>=1.9.0
anndata>=0.9.0
numpy>=1.24.0
pandas>=2.0.0
matplotlib>=3.7.0
seaborn>=0.12.0
leidenalg>=0.10.0
python-igraph>=0.10.0

# Optional: Batch correction
harmonypy>=0.0.6
scanorama>=1.7.3
scvi-tools>=1.0.0
bbknn>=1.6.0

# Optional: Advanced analysis
cellchatpy>=0.1.0
scvelo>=0.3.0
scikit-misc>=0.3.0

# Optional: Spatial transcriptomics
squidpy>=1.3.0
spatialdata>=0.0.14

SKILL.md:220-227:

r
remotes::install_github("satijalab/seurat-wrappers")
remotes::install_github("immunogenomics/harmony")
BiocManager::install("SingleR")
bash
pip install scanpy leidenalg scvi-tools cellchatpy

SKILL.md:294:

bash
pip install -r requirements.txt

scripts/main.py:634:

bash
pip install scanpy leidenalg python-igraph

Technical Analysis

The root dependency file contains no version constraints or integrity hashes. The reference dependency file specifies only lower bounds, so future and potentially incompatible releases remain eligible for installation. The generated documentation similarly directs users to install packages without pinning versions or verifying package artifacts.

The root file requests the distribution named scvi, while the documentation and reference requirements request scvi-tools. The generated Scanpy template imports the Python module scvi, but an import-module name does not establish that the corresponding package distribution should also ...[truncated 2290 chars]

Remediation
View remediation

Remediation Suggestions

  1. Replace the root scvi dependency with the intended and verified distribution, presumed from the other project files to be scvi-tools.
  2. Consolidate the root and reference requirement files into one authoritative dependency specification to prevent package-name and version drift.
  3. Pin every direct dependency to an exact reviewed version, for example package==x.y.z.
  4. Generate and enforce cryptographic hashes using a lock-file workflow such as pip-tools with pip install --require-hashes.
  5. Lock transitive dependencies as well as direct dependencies and regenerate the lock file only through a reviewed update process.
  6. Pin GitHub dependencies to reviewed immutable commit SHAs, for example owner/repository@COMMIT_SHA, rather than mutable default branches.
  7. Separate optional analysis dependencies from the default installation set and install them only when the corresponding feature is requested.
  8. Add automated dependency scanning and package-origin verification to CI.
  9. Recommend installation in an unprivileged, isolated virtual environment or container and explicitly discourage system-wide or administrative installation.
  10. Test the locked dependency set to confirm that import scvi is provided by the selected scvi-tools distribution and that generated templates remain compatible.
Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (35)

Possible Typosquatting: 'scanpy' resembles popular package 'scrapy'

High
Category
Supply Chain
Confidence
70% confidence
Finding

Package name closely resembles a popular package, suggesting possible typosquatting. Attackers publish malicious packages with similar names to trick developers into installing them.

Content

No source excerpt is available for this finding.

Possible Typosquatting: 'scanpy' resembles popular package 'scrapy'

High
Category
Supply Chain
Confidence
70% confidence
Finding

Package name closely resembles a popular package, suggesting possible typosquatting. Attackers publish malicious packages with similar names to trick developers into installing them.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
89% confidence
Finding

The skill documents file-writing behavior by generating analysis scripts and output directories, but it does not declare any explicit tool scope such as allowed-tools or permissions. That creates an authorization gap where an agent may be allowed to write files without a narrowly defined boundary, increasing the risk of unintended or overly broad filesystem access if the skill is invoked in a permissive runtime.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The trigger language is broad enough to activate for many generic bioinformatics or preprocessing requests, which can cause the skill to run in situations the user did not specifically intend. Overbroad activation increases the chance of unnecessary code generation and file creation, and in agentic settings may let this skill influence workflows beyond its narrowly appropriate scope.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The script writes generated files directly to a user-supplied output path and uses mode 'w', which will silently overwrite existing files. In an agent/skill context, this is risky because a caller or prompt-influenced argument could target sensitive or important local files, causing data loss or unintended modification without confirmation.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
95% confidence
Finding

The dependency is specified with a lower bound only, so builds may resolve to different future versions over time. This weakens reproducibility and can unintentionally introduce vulnerable or incompatible releases through the software supply chain, though the file itself does not directly execute malicious behavior.

Content

Scanner excerpt · references/requirements.txt (reported line 1)May include surrounding context.

text
scanpy>=1.9.0
anndata>=0.9.0
numpy>=1.24.0
pandas>=2.0.0

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
95% confidence
Finding

The package version is unpinned and allows any release at or above the minimum, which makes installations non-reproducible and can pull in later vulnerable versions. In a scientific pipeline skill this is primarily a supply-chain hygiene issue rather than an immediately exploitable code flaw.

Content

Scanner excerpt · references/requirements.txt (reported line 2)May include surrounding context.

text
scanpy>=1.9.0
anndata>=0.9.0
numpy>=1.24.0
pandas>=2.0.0
matplotlib>=3.7.0

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
97% confidence
Finding

Using an open-ended constraint for numpy means the installed version is not fixed and could vary across environments, making it impossible to know whether a vulnerable release is being used. This increases supply-chain risk and complicates incident response and reproducibility.

Content

Scanner excerpt · references/requirements.txt (reported line 3)May include surrounding context.

text
scanpy>=1.9.0
anndata>=0.9.0
numpy>=1.24.0
pandas>=2.0.0
matplotlib>=3.7.0
seaborn>=0.12.0

Unverifiable Dependency: numpy has 16 known advisory(ies) (CVE-2014-1859 (Numpy arbitrary file write via symlink attack); CVE-2021-41495 (NumPy NULL Pointer Dereference); CVE-2021-33430 (NumPy Buffer Overflow (Disputed)) +13 more), but the manifest does not pin a version, so it is unknown whether the installed release is affected

Low
Category
Supply Chain
Confidence
90% confidence
Finding

Numpy has known advisories, and because the manifest does not pin an exact version there is no way to verify whether installations will avoid affected releases. The danger here is not that the file proves a vulnerable version is present, but that it prevents meaningful assurance and may allow vulnerable versions into downstream environments.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
96% confidence
Finding

An unpinned pandas dependency allows resolver-selected future releases, which can introduce known or newly disclosed issues without code changes in the skill. The risk is limited here because this file is only a manifest, but it still creates avoidable supply-chain exposure.

Content

Scanner excerpt · references/requirements.txt (reported line 4)May include surrounding context.

text
scanpy>=1.9.0
anndata>=0.9.0
numpy>=1.24.0
pandas>=2.0.0
matplotlib>=3.7.0
seaborn>=0.12.0
leidenalg>=0.10.0

Unverifiable Dependency: pandas has 1 known advisory(ies) (CVE-2020-13091 (** DISPUTED ** pandas through 1.0.3 can unserialize and execute commands from an)), but the manifest does not pin a version, so it is unknown whether the installed release is affected

Low
Category
Supply Chain
Confidence
88% confidence
Finding

Pandas has a known advisory history, and the open-ended version specifier makes it impossible to determine whether an installed version is affected. In this context that is a low-severity but real assurance gap in supply-chain security.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
94% confidence
Finding

The matplotlib requirement is not fixed to a single version, so deployments may drift and consume unreviewed releases. This is a low-severity dependency management weakness that affects integrity and reproducibility more than it enables direct exploitation.

Content

Scanner excerpt · references/requirements.txt (reported line 5)May include surrounding context.

text
anndata>=0.9.0
numpy>=1.24.0
pandas>=2.0.0
matplotlib>=3.7.0
seaborn>=0.12.0
leidenalg>=0.10.0
python-igraph>=0.10.0

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
94% confidence
Finding

The seaborn entry uses a minimum version only, permitting uncontrolled upgrades during installation. That can introduce vulnerable or breaking versions and makes the environment harder to audit consistently.

Content

Scanner excerpt · references/requirements.txt (reported line 6)May include surrounding context.

text
numpy>=1.24.0
pandas>=2.0.0
matplotlib>=3.7.0
seaborn>=0.12.0
leidenalg>=0.10.0
python-igraph>=0.10.0

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
94% confidence
Finding

The leidenalg package is unpinned, allowing future resolver choices that may not have been validated for security or compatibility. In this context it is a low-risk but real supply-chain and reproducibility concern.

Content

Scanner excerpt · references/requirements.txt (reported line 7)May include surrounding context.

text
pandas>=2.0.0
matplotlib>=3.7.0
seaborn>=0.12.0
leidenalg>=0.10.0
python-igraph>=0.10.0

# Optional: Batch correction

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
94% confidence
Finding

python-igraph is declared with a lower bound only, so installations may pick different versions over time. This can expose consumers to unreviewed releases and undermines deterministic builds.

Content

Scanner excerpt · references/requirements.txt (reported line 8)May include surrounding context.

text
matplotlib>=3.7.0
seaborn>=0.12.0
leidenalg>=0.10.0
python-igraph>=0.10.0

# Optional: Batch correction
harmonypy>=0.0.6

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
93% confidence
Finding

The harmonypy dependency is not pinned, which permits varying installed versions and unreviewed upgrades. As an optional package this affects fewer deployments, but it still introduces unnecessary supply-chain uncertainty where used.

Content

Scanner excerpt · references/requirements.txt (reported line 11)May include surrounding context.

text
python-igraph>=0.10.0

# Optional: Batch correction
harmonypy>=0.0.6
scanorama>=1.7.3
scvi-tools>=1.0.0
bbknn>=1.6.0

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
93% confidence
Finding

scanorama is specified with an open-ended version constraint, so future installations may consume different releases with unknown security posture. This is a low-severity dependency hygiene issue, especially for optional components.

Content

Scanner excerpt · references/requirements.txt (reported line 12)May include surrounding context.

text
# Optional: Batch correction
harmonypy>=0.0.6
scanorama>=1.7.3
scvi-tools>=1.0.0
bbknn>=1.6.0

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
93% confidence
Finding

The scvi-tools package is unpinned, allowing resolver-selected versions outside a reviewed baseline. That can silently introduce vulnerable or incompatible releases into environments using this skill.

Content

Scanner excerpt · references/requirements.txt (reported line 13)May include surrounding context.

text
# Optional: Batch correction
harmonypy>=0.0.6
scanorama>=1.7.3
scvi-tools>=1.0.0
bbknn>=1.6.0

# Optional: Advanced analysis

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
93% confidence
Finding

bbknn is declared with a minimum version only, resulting in non-deterministic dependency resolution across installs. While not directly exploitable by itself, it increases supply-chain risk and reduces auditability.

Content

Scanner excerpt · references/requirements.txt (reported line 14)May include surrounding context.

text
harmonypy>=0.0.6
scanorama>=1.7.3
scvi-tools>=1.0.0
bbknn>=1.6.0

# Optional: Advanced analysis
cellchatpy>=0.1.0

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
93% confidence
Finding

cellchatpy is unpinned, so different environments may install different releases with different security properties. This is especially relevant for less common packages where release quality may vary and review coverage may be lower.

Content

Scanner excerpt · references/requirements.txt (reported line 17)May include surrounding context.

text
bbknn>=1.6.0

# Optional: Advanced analysis
cellchatpy>=0.1.0
scvelo>=0.3.0
scikit-misc>=0.3.0

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
93% confidence
Finding

The scvelo version is not fixed, enabling automatic adoption of future releases that may contain vulnerabilities or breaking changes. In this analysis skill that is a low-impact but legitimate supply-chain weakness.

Content

Scanner excerpt · references/requirements.txt (reported line 18)May include surrounding context.

text
# Optional: Advanced analysis
cellchatpy>=0.1.0
scvelo>=0.3.0
scikit-misc>=0.3.0

# Optional: Spatial transcriptomics

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
93% confidence
Finding

scikit-misc is specified as a minimum version rather than an exact one, so installations are not reproducible and may incorporate unreviewed releases. This creates low-level supply-chain risk rather than a direct code execution flaw in the file itself.

Content

Scanner excerpt · references/requirements.txt (reported line 19)May include surrounding context.

text
# Optional: Advanced analysis
cellchatpy>=0.1.0
scvelo>=0.3.0
scikit-misc>=0.3.0

# Optional: Spatial transcriptomics
squidpy>=1.3.0

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
93% confidence
Finding

The squidpy dependency is unpinned and can resolve to different future versions, increasing uncertainty around security and compatibility. Because it is optional, exposure depends on whether users install the spatial analysis extras.

Content

Scanner excerpt · references/requirements.txt (reported line 22)May include surrounding context.

text
scikit-misc>=0.3.0

# Optional: Spatial transcriptomics
squidpy>=1.3.0
spatialdata>=0.0.14

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
93% confidence
Finding

spatialdata is not pinned to an exact release, which allows version drift and potential introduction of vulnerable or incompatible builds. This is a standard low-severity dependency management issue.

Content

Scanner excerpt · references/requirements.txt (reported line 23)May include surrounding context.

text
# Optional: Spatial transcriptomics
squidpy>=1.3.0
spatialdata>=0.0.14

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
95% confidence
Finding

The dependency list leaves bbknn unpinned, so installations may pull different versions over time. That weakens reproducibility and can expose the environment to newly introduced vulnerable or malicious upstream releases through normal dependency resolution.

Content

Scanner excerpt · requirements.txt (reported line 1)May include surrounding context.

text
bbknn
harmonypy
matplotlib
numpy