The Agent Skills Directory

[SAFE]: The skill operates within the standard Elixir and Phoenix development environment, utilizing established tools like mix tasks and Ecto migrations for its primary functions.
[SAFE]: The documentation provides extensive safety warnings and non-negotiable rules for the agent to follow regarding database locks, coordinated schema upgrades, and state reconciliation to prevent data loss or service disruption.
[SAFE]: External dependencies and tools mentioned (such as pgmq, pg_cron, and Supabase CLI) are relevant to the skill's specific purpose of managing a workflow engine and are documented for legitimate developer usage.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection as it scaffolds Elixir code and SQL migrations based on user-provided descriptions and module names.
Ingestion points: User input provided to subcommands such as /pgflow flow, /pgflow job, and /pgflow step (defined in SKILL.md, references/flow.md, and references/step.md).
Boundary markers: The instructions do not define specific delimiters or "ignore" warnings for user-supplied descriptions; the agent is simply told to "derive the steps from that description."
Capability inventory: The skill has the capability to write files to the project directory (lib/), execute shell commands (mix), and perform database DDL operations (ecto.migrate).
Sanitization: There are no explicit instructions for the agent to sanitize or escape user-provided logic or descriptions before interpolating them into scaffolded code or comments.

pgflow