The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted metadata from the JFrog Public Catalog and Stored Packages which is interpolated into shell command arguments and used to guide subsequent diagnostic steps.
Ingestion points: Data enters the context via OneModel GraphQL API responses and Curation API status check results.
Boundary markers: No explicit instruction delimiters or boundary markers are defined in the workflow prompts.
Capability inventory: The skill executes shell commands via the jf CLI, including jf api for REST calls and jf rt dl for binary downloads, and writes temporary files to /tmp.
Sanitization: Uses jq to safely construct JSON payloads and parse responses, mitigating direct command injection risks in the shell environment.
[COMMAND_EXECUTION]: Executes shell commands using the jf (JFrog CLI) and jfs_curation_* tool families. This execution is necessary for the skill's primary functions of repository management and artifact retrieval.
[EXTERNAL_DOWNLOADS]: Fetches package metadata and binary artifacts from the JFrog Platform and Public Catalog. It also retrieves documentation from official JFrog domains (docs.jfrog.com). These operations target known vendor infrastructure.

jfrog-package-curation