The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install a CLI tool from a third-party repository (dicklesworthstone/tap/cass) using Homebrew. This introduces a dependency on an unverified external source not associated with the skill author or a trusted organization.
[COMMAND_EXECUTION]: The skill provides instructions for executing various subcommands of the cass tool. These commands include flags that point to local file system paths and workspaces (e.g., --workspace /path/to/project, cass view /path/to/session.jsonl), enabling the tool to read data from the local environment.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection risk as its primary purpose is to retrieve and display content from previous agent sessions and transcripts.
Ingestion points: Data is ingested from external session files and logs via the search, view, and expand commands in SKILL.md.
Boundary markers: The instructions do not specify any delimiters or safety markers to isolate the retrieved transcript content from the agent's current task instructions.
Capability inventory: The skill allows the execution of shell commands and file system access.
Sanitization: There is no mention of sanitizing or escaping the retrieved session content before it is processed by the agent, meaning malicious instructions found in historical logs could potentially influence the agent's behavior.

cass