The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill retrieves public documentation and upgrade guidance from official vendor domains via the web tool. These references are used for research and citation purposes only.
The skill targets established Splunk documentation sites including help.splunk.com and splunkbase.splunk.com to establish supported upgrade paths and prerequisites.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data supplied by users, such as log files, runbooks, and environment inventories, creating a vulnerability surface for indirect prompt injection. The skill mitigates this risk through explicit defensive prompting and limited tool capabilities.
Ingestion points: Artifacts are ingested as described in the Prerequisites section of SKILL.md, including pasted runbooks, inventories, logs, and retrieved pages.
Boundary markers: The skill contains mandatory instructions in SKILL.md to treat all supplied artifacts as untrusted data and never as instructions, specifically forbidding the agent from following embedded commands.
Capability inventory: The skill's capabilities are limited to the web tool for documentation retrieval and metadata fields; it has no access to shell execution, file system writes, or private network operations.
Sanitization: The instructions in SKILL.md and readiness-contract.md (Smallest safe evidence route) mandate requesting only sanitized fields or bounded artifacts, rather than raw configurations or broad logs.

upgrade-planning-and-execution-readiness