The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external data from PingOne DaVinci execution logs, audit events, and flow definitions, which presents a surface for instructions embedded in logs to influence agent behavior.
Ingestion points: references/playbooks/debug-davinci-flow.md details the retrieval of execution events, flow definitions, and audit events from the PingOne environment.
Boundary markers: While assets/templates/EXECUTION_ANALYSIS_REPORT.md enforces a markdown structure for the final output, there are no explicit boundary markers or instructions to ignore embedded commands within the raw ingested data.
Capability inventory: The skill uses pingcli and PingOne APIs (documented in references/cli.md) to perform environment listing, user searching, and raw API requests against production identity infrastructure.
Sanitization: references/troubleshooting.md provides instructions to minimize PII and avoid persisting data beyond the current session, but does not specify technical filtering of the ingested content for injection patterns.
[COMMAND_EXECUTION]: The skill relies on executing the pingcli tool, including raw API interactions via pingcli pingone api, to troubleshoot flow failures. These commands are executed within the user's authorized environment to manage identity resources.
[EXTERNAL_DOWNLOADS]: The skill references documentation and machine-readable index files from the vendor's official developer portal at developer.pingidentity.com. These resources are used to supplement the agent's knowledge of the PingOne API and setup options.

davinci-flow-engineering