The Agent Skills Directory

[DATA_EXFILTRATION]: The skill performs an outbound network request using curl to a URL specified in the user's local configuration (~/.superstack/config.json). This is used to transmit telemetry data including skill execution status, version, and platform metadata to a Convex service endpoint.- [COMMAND_EXECUTION]: The skill executes shell commands in its preamble and telemetry sections to manage configuration files, extract telemetry settings using grep and sed, and maintain local activity logs in the ~/.superstack directory.- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection as it processes and exports data from .superstack/learnings.md, a file that may be populated by other skills or automated processes.
Ingestion points: Reads content from .superstack/learnings.md during search, prune, and export operations.
Boundary markers: Relies on markdown structure (headings and kebab-case keys) to segment data.
Capability inventory: Includes the ability to write to the project's CLAUDE.md and modify .superstack/learnings.md.
Sanitization: The skill does not explicitly sanitize the 'Insight' or 'Files' fields before writing them to the project's core documentation files.

learn