The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the pscale CLI tool to perform administrative database operations. It includes a utility script, scripts/create-branch-for-mr.sh, which uses regular expression validation to prevent command injection by ensuring database and branch names contain only safe characters.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources that could be manipulated to include malicious instructions for the agent.
Ingestion points: The skill instructions direct the agent to query and process database logs (pscale logs), schema definitions (pscale branch schema), and query pattern reports (pscale branch query-patterns download) from the PlanetScale environment (SKILL.md).
Boundary markers: Absent; there are no specific instructions to use delimiters or ignore instructions found within the retrieved database data.
Capability inventory: The skill allows the agent to modify database infrastructure, including creating, deleting, and promoting branches, and executing SQL via pscale shell (SKILL.md).
Sanitization: Partial; the skill includes a specific warning for the agent to avoid exposing sensitive information such as SQL text, usernames, or addresses from log messages, though it does not provide technical filtering for prompt injection (SKILL.md).
[DATA_EXFILTRATION]: The skill retrieves and exposes database-related information, such as PostgreSQL/Neki logs and query pattern reports, to the agent's context. This data may contain sensitive application or server-side information.
Evidence: Use of pscale logs and pscale branch query-patterns download to retrieve environment data (SKILL.md).

pscale-branch