Back to skill

Security audit

LYGO Champion: cryptosophia soulforger

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a low-impact persona helper, but it tells users to run an unpinned remote installer for a successor skill, which should be reviewed first.

Install only if you are comfortable with the CRYPTOSOPHIA persona behavior and the legacy-successor relationship. Before following the upgrade command, verify the `clawhub` package source, prefer a pinned reviewed version, and separately review the successor skill being installed.

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:9
Finding
Unpinned Third-Party CLI Execution Through npx## Vulnerability Details **File Location**: `SKILL.md:9` **Vulnerability Type**: Unpinned third-party dependency execution **Risk Level**: Medium **Vulnerable Code Snippet**: ```markdown > `npx clawhub@latest install deepseekoracle/lygo-champion-council` ``` ### Technical Analysis The documented upgrade procedure instructs users to execute the mutable `latest` release of the third-party `clawhub` npm package through `npx`. Depending on the local npm configuration and cache state, `npx` may retrieve the package from a remote registry and execute its CLI with the invoking user's privileges. Because the command does not pin an audited package version or verify a package integrity digest, the code executed in the future may differ from the code available when this skill was reviewed. The command also installs a successor skill whose contents are outside the audited project. A compromised publisher account, package release, registry path, or other supply-chain component could therefore cause attacker-controlled code to run when a user follows the documented instruction. ### Attack Path 1. An attacker compromises the `clawhub` package publisher, its release process, or another relevant supply-chain component. 2. The attacker publishes a malicious release under the mutable `latest` tag. 3. A user follows the upgrade command documented in `SKILL.md`. 4. `npx` retrieves or resolves the malicious package release. 5. The package CLI executes with the permissions of the user running the command. 6. The malicious process may access or modify resources available to that user and may install an unreviewed successor skill. ### Impact Assessment Successful exploitation would provide code execution with the privileges of the invoking user. The accessible scope could include project files, user-readable local data, environment variables exposed to the process, and files writable by that account. The malicious package could also modi ...[truncated 257 chars]
Remediation
## Remediation Suggestions - Replace the mutable `@latest` selector with an exact, reviewed package version. - Verify the package against a documented integrity digest or trusted lockfile before execution. - Document the expected package publisher, registry source, version, and integrity value. - Review the successor skill and its installation behavior before recommending it to users. - Avoid presenting remote package execution as a copy-and-paste upgrade step. Prefer a staged process that downloads or resolves the package, verifies its provenance and integrity, and only then executes it. - Re-audit and deliberately update the pinned version and integrity value whenever an upgrade is required.
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (5)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding
The declared description says this skill is a deprecated legacy slug that should be replaced by another skill. That implies a metadata-only or redirect-like purpose. The supplied code instead actively performs integrity and configuration checks over local files and reference content. This is a materially different primary purpose: operational self-validation versus deprecation notice. No dangerous external actions are present, but the behavior is still inconsistent with the declared description.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding
The declared purpose says this is a deprecated legacy slug with no triggers or permissions, implying metadata/alias behavior rather than functional code. The actual code performs a concrete operation: accessing a local JSON reference file and printing a specific hash value. That is a materially different primary purpose from a deprecated slug/redirect description, so this should be flagged as a mismatch.

Lp3

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding
Without declared permissions the skill's intent is opaque and cannot be validated.

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding
The installation command uses `npx clawhub@latest`, which pulls the latest published package at execution time rather than a pinned, audited version. If the upstream package is compromised or a breaking/malicious release is published, users following the instruction could execute untrusted code on their system.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The activation flow is phrased as a user ritual with broad, natural-language triggers like sharing symbolic content, speaking an invocation, and then co-creating. In an agent setting, vague activation criteria can cause accidental persona switching or unintended behavioral changes when ordinary user input matches the described pattern, which may bypass expected control boundaries or confuse downstream logic.

Static analysis

No suspicious patterns detected.