The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill focuses on analyzing error logs, stack traces, and code snippets provided as external input. This workflow presents an attack surface where malicious instructions could be embedded in the data processed by the agent. Ingestion points: Untrusted data enters the agent's context through user-pasted error messages, log files, and code snippets used for diagnosis. Boundary markers: The skill's instructions do not specify the use of delimiters or clear directives to ignore instructions that may be contained within the analyzed data. Capability inventory: Across its files, the skill provides and encourages the use of numerous shell commands (npm, pip, git, curl, systemctl, chmod, chown), file system writes (recording solutions in .claude/error-solutions/), and network diagnostic tools. Sanitization: There is no mention of sanitizing or validating input data before it is used for pattern matching or remediation steps.
[COMMAND_EXECUTION]: The skill contains a vast library of shell commands for debugging across multiple languages (Node.js, Python, Go, Rust) and environments (Docker, Git, Databases). These include powerful operations such as modifying file permissions (chmod), changing file ownership (chown), and executing commands with administrative privileges (sudo), which are presented as common troubleshooting steps.

Error Resolver