The Agent Skills Directory

[COMMAND_EXECUTION]: Path Traversal Vulnerability. The scripts/unpack.sh script is vulnerable to path traversal because it parses file paths from markdown headers and uses them directly in file system operations without validation. A malicious context pack can use directory traversal sequences like ../ to write or overwrite files outside the intended destination, such as system configuration files or SSH keys.
[PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill aggregates content from arbitrary files into a single context window for an LLM.
Ingestion points: Files are read from the local filesystem or extracted from ZIP archives in scripts/pack.sh.
Boundary markers: Content is delimited by markdown headers and triple backticks, but no instructions are provided to the agent to ignore embedded commands.
Capability inventory: The skill has the capability to write files and access the system clipboard.
Sanitization: No sanitization or escaping of the packed file content is performed.
[COMMAND_EXECUTION]: Deceptive Documentation. The skill documentation describes advanced features such as an interactive TUI and a web UI (scripts/pack-web.sh) that are not implemented in the provided shell scripts. This can lead to confusion and represents a failure to provide the promised functionality.

context-pack