The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands, specifically the JFrog CLI (jf api) and curl, to query and modify platform settings. These commands are dynamically constructed using data extracted from the user's manifest file through yq and jq.
[DATA_EXFILTRATION]: The skill performs network operations to interact with JFrog API endpoints and external state backends (Artifactory or Git) to fetch and store configuration manifests. These operations are restricted to the environments configured by the user for state management.
[PROMPT_INJECTION]: The skill ingests untrusted data via YAML manifests, creating a surface for indirect prompt injection where manifest values are used in subsequent CLI commands.
Ingestion points: Desired-state manifest YAML file processed in Phase 1 and Phase 2.
Boundary markers: The skill implements a mandatory 'Present Diff' phase where all planned actions must be reviewed and approved by the user before execution.
Capability inventory: Shell command execution via jf and curl, network access to JFrog services, and the ability to trigger auxiliary configuration skills.
Sanitization: Data is parsed using structured utilities (yq, jq) which helps prevent raw content from being treated as instructions.

jfrog-reconcile-manifest