The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because its primary function is to ingest and analyze untrusted external data, such as stack traces, error logs, and user reports.
Ingestion points: Data enters the agent's context through user inputs and the output of tools like Read and Grep during the systematic triage process defined in SKILL.md and checklists/systematic-debugging-checklist.md.
Boundary markers: The instructions do not define clear delimiters or include warnings to ignore instructions that might be embedded within the logs or traces being analyzed.
Capability inventory: The skill has access to a powerful toolset including Bash, Write, and MultiEdit, which increases the potential impact of an successful indirect injection.
Sanitization: There are no documented procedures for sanitizing or escaping external content before the agent processes it.
[COMMAND_EXECUTION]: The skill includes instructions and templates that involve executing shell commands and modifying project configuration files.
Evidence: In reference/error-patterns-database.md, the skill provides templates for executing pip install and appending to requirements.txt to fix import errors. While appropriate for debugging, these actions could be leveraged to perform unauthorized package installations if the agent's logic is influenced by malicious input.

smart-debugging