The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses a pattern in Phase 1 where JSON data fetched from crt.sh is piped directly into a Python interpreter via python3 -c. While the Python logic itself is provided in the skill, piping network-retrieved data to an interpreter is a high-risk pattern that can be exploited if the source service is compromised or if the data triggers interpreter vulnerabilities.
[COMMAND_EXECUTION]: The skill extensively uses shell commands including curl, grep, and tee to automate reconnaissance and manipulate data. This includes looping through targets and executing complex command chains.
[DATA_EXFILTRATION]: The instructions focus on identifying and extracting PII (Personally Identifiable Information) from targets. It provides specific regular expressions to harvest emails, phone numbers, addresses, and credit-related data from server logs and directory listings.
[EXTERNAL_DOWNLOADS]: The skill performs numerous unauthenticated network requests to download content from target domains and external certificate transparency logs.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its processing of untrusted external data.
Ingestion points: Phase 2 downloads and processes debug.log files from target WordPress sites.
Boundary markers: No delimiters or ignore instructions are present to protect the agent from malicious commands embedded in the target logs.
Capability inventory: The skill has shell execution (curl, bash), file system access (/tmp/), and network operation capabilities.
Sanitization: No sanitization or validation of the downloaded log content is performed before the agent processes the file.

recon-plumbing