The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Instructions are provided to install the healthmd CLI using Homebrew from the author's tap (CodyBontecou/tap/healthmd) and to clone the source code from the author's GitHub repository (github.com/CodyBontecou/health-md).
[REMOTE_CODE_EXECUTION]: The skill guides the user to clone a remote repository and execute a build/installation command (cargo install --locked --path crates/healthmd-cli), which results in the execution of code downloaded from a remote source.
[COMMAND_EXECUTION]: The skill makes extensive use of shell commands to install and operate the CLI tool, including package management (brew install), version control (git clone, git checkout), and tool-specific operations (healthmd export, healthmd query).
[DATA_EXFILTRATION]: The skill is designed to extract sensitive health data from an iPhone and export it to local files or stream it to the agent's context. This represents the intended behavior of the tool but constitutes handling of highly sensitive information.
[INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it ingests health data from an external mobile device into the agent's context.
Ingestion points: Health data is ingested through the healthmd export and healthmd query commands described in SKILL.md.
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following potential instructions embedded within the health data records.
Capability inventory: The skill possesses the ability to execute shell commands and write to the local file system using the CLI tool.
Sanitization: The documentation mentions validation of data integrity and manifests, but does not specify sanitization of the health data content for embedded natural language instructions.

healthmd-cli