The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process and manage data such as local files, conversation histories, and GitHub repository content which could potentially contain malicious instructions.
Ingestion points: Data enters the agent's context through subcommands like helmor files and helmor session as referenced in SKILL.md.
Boundary markers: The provided instructions do not include the use of explicit delimiters or protective prompts to mitigate instructions embedded in the ingested data.
Capability inventory: The skill can execute various CLI subcommands, perform file system operations (read/write/stage), and dispatch tasks to other agents via the helmor send command.
Sanitization: There is no evidence of sanitization or validation protocols for data retrieved from external repositories or session logs.
[DYNAMIC_EXECUTION]: The skill includes functionality to inspect and manage scripts within a repository environment.
Evidence: The helmor scripts command group is used to handle repo-level setup, run, and archive scripts, as detailed in SKILL.md and references/helmor-help.md.
Context: This represents a script management capability common in development tools, though it allows the agent to interact with executable content within the user's workspace.

helmor-cli