The Agent Skills Directory

[PERSISTENCE]: The skill instructs the agent to install native hooks at the user level via the forge hooks install --global command. These hooks are designed to trigger during specific lifecycle events, representing a persistence mechanism that allows the tool to maintain influence over the agent's operating environment across different sessions.
[COMMAND_EXECUTION]: The skill promotes the use of the Bash tool with flags like --force and --yes (e.g., in forge reset --hard --force and forge setup --yes). These commands are specifically designed to bypass interactive safety prompts and perform destructive operations or configuration changes without manual confirmation, which could be exploited to cause unintended data loss.
[INDIRECT_PROMPT_INJECTION]: The skill functions by interpolating user-provided inputs (such as harness names, directory paths, and configuration profiles) into shell commands executed via the Bash tool, creating a surface for injection.
Ingestion points: Parameters supplied in user queries for initialization and setup commands in SKILL.md.
Boundary markers: None identified to delimit instructions from user-supplied data within the command strings.
Capability inventory: The skill has access to Bash, Read, Grep, and Glob tools as specified in the SKILL.md frontmatter.
Sanitization: There are no documented sanitization or validation routines to prevent command injection when user input is concatenated into the command-line arguments.

setup