The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill implements a dangerous pattern where data fetched from a remote, untrusted target via curl is piped directly into a python3 command for processing. Examples include piping output from crt.sh and the target's /wp-json/ endpoint directly to Python script blocks in SKILL.md.
[COMMAND_EXECUTION]: The skill is comprised of numerous shell commands that execute network probes, file system operations (using tee and -o), and various data processing utilities like grep, jq, and wc.
[DATA_EXFILTRATION]: The skill's primary purpose is to identify and extract sensitive data from property management sites. It contains specific logic to search for PII such as Social Security Numbers, bank account information, and lease agreements within exposed debug.log files or directory listings.
[EXTERNAL_DOWNLOADS]: The skill fetches certificate information from the well-known crt.sh service and retrieves configuration data from various target-controlled endpoints.
[PROMPT_INJECTION]: The skill is vulnerable to Indirect Prompt Injection because it ingests untrusted data from external targets (e.g., REST API responses and debug logs) without boundary markers or sanitization. This data is then processed by tools with command execution and file-write capabilities.

recon-property-management