T09 · Insecure Skill Coding Practices
Warning
- Location
scripts/ontology.py:28- Finding
Missing Schema Enforcement Allows Plaintext Credential Storage and Invalid Graph Records
- Content
View full analysis
- Remediation
View remediation
Security audit
Agent Ontology
Security checks for vulnerabilities and agentic risk
Overview
This is a purposeful local memory skill, but it needs Review because persistent graph writes are broadly scoped and the promised validation and secret protections are not actually enforced.
Install only if you are comfortable with a persistent local memory file. Avoid storing passwords, API keys, tokens, private messages, or regulated personal data in it until schema validation, secret-field rejection, file permission hardening, and deletion/update controls are implemented.
Vulnerability Patterns
- Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
- 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)
Vulnerability Patterns
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
- Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
- MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (4)
Credential Access
High
- Category
- Privilege Escalation
- Confidence
- 70% confidence
- Finding
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
- Content
forbidden_properties: [password, secret, token, key, api_key] properties: service: string secret_ref: string # Reference to secret store (e.g., "keychain:github-token") expires: datetime? scope: string[]?
text
Vague Triggers
Medium
- Category
- Not specified by scanner
- Confidence
- 89% confidence
- Finding
The activation examples are very broad conversational phrases such as "Remember that..." and "What do I know about X?", which can overlap with normal user dialogue and cause the skill to activate unexpectedly. In a skill that performs knowledge-graph writes and updates, over-broad triggering increases the chance of unintended persistence, incorrect graph mutations, or disclosure of stored information when the user did not explicitly intend to use this capability.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 96% confidence
- Finding
The skill explicitly stores information in a persistent append-only graph file but does not warn users that their inputs may be retained. This creates a privacy and data-governance risk because users may disclose personal, sensitive, or contextual information under the assumption it is ephemeral, and append-only storage also makes later deletion or correction harder.
- Content
Intent-Code Divergence
Medium
- Category
- Not specified by scanner
- Confidence
- 96% confidence
- Finding
The module docstring states this CLI provides 'CRUD operations on typed knowledge graph,' which implies create, read, update, and delete support. In the actual code, commands only implement create/read-style operations and relationship creation; there are no update or delete commands or corresponding code paths.
- Content