The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill retrieves bookmark titles, descriptions, and notes from the Linkding API. Because these fields contain user-controlled data, they could potentially host malicious instructions designed to influence the agent's behavior when it reads or searches bookmarks.
Ingestion points: Data enters the agent's context through scripts/linkding-api.sh when fetching bookmarks, tags, or bundles via API GET requests.
Boundary markers: The skill does not define clear delimiters or use explicit instructions to ignore embedded commands within the bookmark data.
Capability inventory: The skill can execute shell commands via curl and jq and has network access to the user's Linkding instance.
Sanitization: While the script uses jq to safely encode data sent to the API, there is no explicit sanitization of the data retrieved from the API before it is processed by the agent.
[COMMAND_EXECUTION]: The skill uses a bash script (scripts/linkding-api.sh) that invokes curl and jq to interact with the API. The documentation also includes several complex shell workflows for tasks like bulk archiving and duplicate detection.

linkding