The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on executing shell commands through npx to interact with the knowcards CLI tool for querying, proposing, and checking the status of local cards.
[EXTERNAL_DOWNLOADS]: The skill invokes npx knowcards, which initiates a download and execution of the knowcards package from the npm registry if it is not already cached locally.
[INDIRECT_PROMPT_INJECTION]: The skill's protocol creates a vulnerability surface by instructing the agent to treat information retrieved from external knowledge cards as "trusted memory" and to prioritize these cards over other repository documentation like READMEs.
Ingestion points: Data returned by the npx knowcards query command.
Boundary markers: Absent; the instructions do not provide delimiters or warnings to ignore potentially malicious instructions embedded within the cards.
Capability inventory: The agent has permissions to execute shell commands (npx) and write to the filesystem (via the propose command).
Sanitization: Absent; the skill lacks mechanisms to filter, validate, or sanitize the content of the cards before processing them as trusted facts.

knowcards