Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed project-memory helper that writes local project notes and safety checklists, with no evidence of hidden execution, exfiltration, or destructive behavior.
Install this only for projects where you want durable AI_DOC notes and task-memory files. Avoid letting the agent write secrets, credentials, private customer data, or sensitive code excerpts into those logs, and periodically review or delete AI_DOC content if it should not remain in the repository.
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
# NEVER execute these:
rm -rf /
mkfs.*
fdisk
format
YARA rule matched a known malware signature (reverse shell, backdoor, ransomware, C2 framework, or info stealer).
ble and follows single responsibility
# NEVER execute these:
rm -rf /
mkfs.*
fdisk
format
dd if=/dev/zero
Rules:
/etc, /usr, /bin, etc.).git/ directory directlytrash instead of rm when availablesudoThe activation conditions are broad enough to trigger on ordinary development discussions such as 'starting a project' or 'complex multi-file projects'. Because the skill imposes mandatory file reads/writes and persistent logging once activated, unintended invocation can cause unnecessary context capture, filesystem changes, and retention of user/project data without clear user intent.
The skill instructs the agent to create persistent per-task memory files for each interaction and to record user intent, cognitive anchors, discoveries, and related context links. This creates a durable natural-language log that can capture sensitive source code details, secrets, internal architecture, or private user information, increasing the risk of later disclosure or misuse.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
**Rules:**
- Never modify system directories (`/etc`, `/usr`, `/bin`, etc.)
- Never operate on `.git/` directory directly
- Never overwrite files without confirmation
- Always use `trash` instead of `rm` when available
### Network Data - FORBIDDEN
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
**Rules:**
- Never modify system directories (`/etc`, `/usr`, `/bin`, etc.)
- Never operate on `.git/` directory directly
- Never overwrite files without confirmation
- Always use `trash` instead of `rm` when available
### Network Data - FORBIDDEN
No suspicious patterns detected.