The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to perform validation by executing shell commands such as bin/synapseq -test, synapseq -test, and go run ./cmd/synapseq. The target file path for these commands is determined by the agent and can be influenced by user input (requested path). If the agent fails to strictly sanitize or escape this path, it could lead to command injection.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests untrusted user descriptions and reads existing sequence files to generate new content, which it then writes to the filesystem and validates via shell commands.
[INDIRECT_PROMPT_INJECTION]: Ingestion points: User-provided task descriptions and existing .spsq or .spsc files (referenced in SKILL.md).
[INDIRECT_PROMPT_INJECTION]: Boundary markers: No explicit delimiters or instructions are provided to isolate untrusted content from the instructions.
[INDIRECT_PROMPT_INJECTION]: Capability inventory: The skill can write new files to the filesystem and execute shell commands for validation.
[INDIRECT_PROMPT_INJECTION]: Sanitization: The instructions lack specific requirements for sanitizing user-provided paths or content before they are used in file operations or shell commands.

create-spsq