First Seen
Feb 18, 2026
Last Scanned
Feb 22, 2026
Findings
4
Score
69/100
Findings (4)
Detects DNS-based data exfiltration techniques
DNS exfil Limit clipboard access to write-only operations or require explicit user confirmation before reading clipboard contents. Block transmission of clipboard data to external endpoints.
Likely FP if the skill is a clipboard manager whose documented purpose is to sync clipboard content with the user's own devices.
Benign heading "'@openai/agents:*'\n;\n// Verbose logging\n..." followed by dangerous content (category: credential_access)
Hardcoded secrets -- API keys (AWS, GitHub, Stripe, OpenAI, Slack), tokens, private keys, connection strings, passwords Ensure section headings accurately reflect the content that follows. Remove headings that could mislead an LLM into treating content differently than intended.
Likely FP if the heading mismatch is due to inconsistent markdown formatting or a benign section title that happens to contain keywords like system or config.
Detects subprocess calls with shell=True which enables shell injection
shell=True Replace shell=True with shell=False and pass command arguments as a list. Validate and sanitize all inputs before passing to the shell.
Likely FP if the match is in documentation describing how subprocess works, or in a comment explaining shell risks rather than actual code.
Detects skills that include unscoped Bash in their allowed tools list (not Bash(cmd:*) scoped)
allowed-tools: Read, Glob, Grep, Bash
Scope the Bash tool to specific commands using allowedTools patterns (e.g., Bash(git *) instead of bare Bash). Remove blanket Bash access from allowed_tools lists.
Likely FP if the Bash entry in allowed_tools is part of a constrained configuration that limits commands elsewhere (e.g., via system prompt restrictions).