The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection because it processes untrusted data from command logs and configuration files. \n
Ingestion points: Captures failing command outputs and error messages as defined in the workflow in SKILL.md. \n
Boundary markers: Absent; no specific instructions are provided to use delimiters or ignore embedded instructions when processing external error logs. \n
Capability inventory: The skill invokes mise CLI commands including trust, install, and use as listed in SKILL.md. \n
Sanitization: Instructions in SKILL.md explicitly require redacting secrets and sensitive environment details before processing. \n- [COMMAND_EXECUTION]: The skill executes the mise command-line utility to perform diagnostic and configuration tasks. Evidence of this intended behavior is found in SKILL.md and references/contract.yaml. \n- [EXTERNAL_DOWNLOADS]: The skill uses mise install and mise upgrade to fetch runtime environments from official and well-known registries. This is a standard and expected operation for the mise toolchain as described in SKILL.md.

fix-mise