The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill describes a /wizard utility that generates interactive bash scripts to automate infrastructure provisioning, credential setup, and environment configuration.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a router for processing external and potentially untrusted data, such as bug reports and feature requests, which can influence subsequent engineering actions.
Ingestion points: Bug reports, incoming feature requests, and external project documentation processed via the /triage and /grill-with-docs workflows as described in SKILL.md.
Boundary markers: The instructions mention "context hygiene" and the use of /compact for summarization, but they do not define specific delimiters or instructions to ignore embedded commands in external data.
Capability inventory: The orchestration flow utilizes tools with significant system capabilities, including file creation and modification (/to-spec, /to-tickets, /implement), test execution (/tdd), and merge conflict resolution.
Sanitization: There is no evidence of content sanitization or validation for the data ingested during the triaging or discovery phases.
[COMMAND_EXECUTION]: The skill documentation includes instructions for executing various system and workflow commands, including a mandatory /setup-matt-pocock-skills precondition.

ask-silva