Back to skill

Security audit

Kannaka Quantum

Security checks for vulnerabilities and agentic risk

Overview

The skill is not deceptive, but it asks agents to install an unpinned package and can provision paid compute, inject API keys, and run autonomous remote agents over SSH without enough scoping guidance.

Review this carefully before installing. Use it only in an isolated environment with scoped, revocable qBraid/OpenQuantum/API credentials. Do not allow paid compute, SSH setup, API-key injection, or autonomous remote-agent launch unless you explicitly intend that action and have a plan to stop compute and remove secrets afterward.

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:17
Finding

Unpinned Third-Party Executable Dependency

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, line 17
Vulnerability Type: Unpinned third-party package installation
Risk Level: Medium

Vulnerable Code

markdown
If neither is available, install the package: `pip install kannaka-quantum` (or `pip install -e .` from the repo). Python ≥ 3.10. If a spawned process can't find Python, set `KANNAKA_QUANTUM_PYTHON` to the interpreter path.

Technical Analysis

The skill instructs the agent to install and execute kannaka-quantum from pip without specifying an audited version, package hash, lockfile, or trusted package index. Consequently, the code installed at execution time may differ from the code that existed when the skill was reviewed.

The project contains only SKILL.md; it does not include the package implementation needed to verify the dependency's installation hooks, runtime behavior, network operations, credential handling, or command execution. Although the audit found no evidence that the named package is malicious, relying on an unpinned executable dependency creates a supply-chain weakness.

The documented package is expected to interact with qBraid and OpenQuantum credentials and can manage paid compute and SSH-based remote agents. A compromised or unexpectedly changed package release would therefore execute in a security-sensitive context.

Attack Path

  1. The skill is invoked in an environment where the kannaka-quantum MCP server and CLI are unavailable.
  2. Following SKILL.md, the agent runs pip install kannaka-quantum.
  3. pip resolves the current package release from its configured index rather than a reviewed, immutable artifact.
  4. An attacker who has compromised the package, its publisher account, the package index, or the dependency-resolution path supplies malicious installation or runtime code.
  5. The code executes with the permissions of the user running pip or the installed CLI.
  6. When the CLI is subsequently used, malicious code may access crede ...[truncated 891 chars]
Remediation
View remediation

Remediation Suggestions

  1. Pin the dependency to a specifically reviewed version, for example with an exact == constraint.
  2. Require cryptographic hashes through a locked requirements file and install with pip install --require-hashes.
  3. State the canonical package repository and trusted package index explicitly.
  4. Review the pinned package and its complete transitive dependency graph before deployment.
  5. Install the dependency in an isolated virtual environment or container under a non-privileged account.
  6. Disable unnecessary access to local credential files and environment variables during installation.
  7. Use narrowly scoped, short-lived qBraid and OpenQuantum credentials at runtime.
  8. Separate package installation from authenticated execution so installation hooks cannot access production credentials.
  9. Prefer bundled, reviewable source or a verified immutable artifact when practical.
  10. Add dependency-integrity monitoring and require manual review before updating the pinned version.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill description is overly broad and bundles low-risk quantum simulation with high-risk actions like provisioning paid compute and launching autonomous remote coding agents over SSH. This ambiguity can cause the agent to invoke remote execution or spend-enabled workflows in response to vague user requests, increasing the chance of unauthorized code execution, credential use, or unexpected billing.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill advertises autonomous remote agent setup over SSH, including API key injection and remote command/control, without an explicit security warning about credential exposure, persistence on remote hosts, or the risks of autonomous execution. In this context, users or downstream agents may treat these operations as routine setup rather than sensitive actions that can lead to remote compromise, secret leakage, or unbounded actions on provisioned infrastructure.

Content

No source excerpt is available for this finding.