The Agent Skills Directory

[DATA_EXFILTRATION]: The skill reads the local configuration file ~/.superstack/config.json to obtain a telemetry URL and sends usage statistics and system metadata (OS and architecture via uname) to this endpoint using curl. The skill includes a consent mechanism that prompts the user before enabling this telemetry.
[COMMAND_EXECUTION]: The skill executes bash scripts during the preamble and conclusion of the workflow to handle telemetry logging, session management, and configuration updates. These scripts interact with local configuration files and the environment.
[INDIRECT_PROMPT_INJECTION]: The skill processes project insights stored in .superstack/learnings.md, which serves as a vulnerability surface for indirect injection.
Ingestion points: The contents of .superstack/learnings.md are read during search, prune, and export operations (SKILL.md).
Boundary markers: No specific boundary markers or instruction-ignoring delimiters are defined for the ingested learning content.
Capability inventory: The skill can execute bash commands, perform network requests via curl, check file existence via globbing, and write to project files such as CLAUDE.md.
Sanitization: The skill does not explicitly mention sanitizing or escaping the text of the insights before they are processed, presented to the user, or exported to other files.

learn