Back to skill

Security audit

Thoth Pro

Security checks for vulnerabilities and agentic risk

Overview

This documentation skill does mostly what it advertises, but it exposes broad project contents in command output and asks users to modify the system Python environment.

Review this before installing on proprietary or secret-bearing repositories. Prefer running it in a disposable virtual environment, set INJECT_DOCSTRINGS=no unless you want source edits, and avoid letting it print raw code or git history into shared logs or remote agent transcripts.

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
SKILL.md:43
Finding

Unpinned Dependency Installation into the System Python Environment

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 43-47
Vulnerability Type: Unpinned third-party dependency installation with system package protections bypassed
Risk Level: Medium

Vulnerable Code Snippet:

markdown
## Step 1 -- Install

```bash
pip3 install rich --break-system-packages --quiet
text

### Technical Analysis

The installation command retrieves `rich` without specifying a reviewed version or verifying an integrity hash. Consequently, the installed artifact can change between executions even though the audited Skill remains unchanged.

The `--break-system-packages` option explicitly bypasses protections for externally managed Python environments. This permits pip to modify the system Python installation and can overwrite or create conflicts with packages used by other applications. This exceeds the safer approach of installing the dependency into a dedicated virtual environment.

The Skill subsequently imports modules from `rich`. If a compromised or otherwise malicious release is installed, attacker-controlled Python code could execute when those modules are imported. Although the package name is not an apparent typosquat and the command uses the normal pip package source configuration, the absence of version and integrity controls leaves a supply-chain exposure.

### Attack Path

1. An attacker compromises the upstream package, its maintainer account, the configured package index, or the dependency resolution path.
2. The attacker publishes or serves a malicious version under the dependency name.
3. A user follows Step 1 and runs the unpinned pip command.
4. Pip resolves and installs the attacker-controlled artifact into the externally managed Python environment because `--break-system-packages` bypasses the normal protection.
5. The Skill executes Step 2, which imports `rich.console`, `rich.panel`, `rich.table`, and other package modules.
6. Malicious package code executes w
...[truncated 698 chars]
Remediation
View remediation

Remediation Suggestions

  1. Create and use a dedicated virtual environment instead of modifying the system Python environment.
  2. Remove the --break-system-packages option.
  3. Pin rich to a specifically reviewed version.
  4. Use a lock file or requirements file containing cryptographic hashes and install with --require-hashes.
  5. Configure pip to use a trusted, explicitly defined package index where appropriate.
  6. Review dependency updates before changing the pinned version.
  7. Consider avoiding the third-party dependency entirely if the output formatting can be implemented with the Python standard library.

Example hardened installation:

bash
python3 -m venv .thoth-venv
.thoth-venv/bin/python -m pip install --require-hashes -r requirements.lock

The lock file should pin the reviewed rich version and all transitive dependencies with approved SHA-256 hashes.

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill enumerates project files, reads up to 5000 characters from each of up to 30 code files, and prints that content to stdout, then also prints recent git commit history. This creates unnecessary exposure of potentially sensitive source code, secrets, internal URLs, credentials, or proprietary history outside the minimum needed to generate local documentation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill reads broad portions of the repository and recent git history, then displays them, but the description does not clearly disclose this data exposure behavior. In the context of proprietary or secret-bearing repositories, this can leak sensitive implementation details, embedded credentials, and internal commit metadata to logs or upstream tooling.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill reads broad portions of the repository and recent git history, then displays them, but the description does not clearly disclose this data exposure behavior. In the context of proprietary or secret-bearing repositories, this can leak sensitive implementation details, embedded credentials, and internal commit metadata to logs or upstream tooling.

Content

No source excerpt is available for this finding.