The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes instructions for the agent to execute a shell command to perform a daily version check. This command utilizes a local timestamp file in the user's home directory (~/.claude/.foundry-version-checked) to implement rate-limiting and uses curl to retrieve current version strings from the vendor's domain (foundry.thehorizonfoundry.com).
[EXTERNAL_DOWNLOADS]: The skill is configured to fetch a plan template from a public GitHub repository associated with the author and performs network requests to the vendor's API for version synchronization. These resources are hosted on infrastructure controlled by the skill's author.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local repository environment, including git commit messages and branch names, to generate plan summaries. This creates a surface for indirect prompt injection. However, the skill mitigates this risk by employing structured markdown templates and requiring human confirmation for key decisions and plan indexing.
[DYNAMIC_EXECUTION]: The skill uses a multi-part shell script for version management and rate-limiting. This script is static within the skill instructions and does not incorporate untrusted external input into the execution flow.

phase-plan