The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of official command-line tools from well-known technology providers.
Evidence: The skill instructs the agent to download and execute the Railway CLI installer from railway.com/install.sh and refers to official package registries for vercel and supabase CLI tools.
[COMMAND_EXECUTION]: The skill executes local shell scripts included in its package to perform read-only discovery of the project's state.
Evidence: SKILL.md executes scripts/scenario-detect.sh to identify the project type and scripts/preview-makefile-wipe.sh to identify existing Makefile scaffolding for safe replacement.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it ingests project-specific data to customize generated content.
Ingestion points: references/agents-md-update.md and SKILL.md read package.json, Cargo.toml, and AGENTS.md to extract project names and existing documentation prose.
Boundary markers: The skill uses specific Markdown headers (e.g., ## Make targets) as delimiters to manage generated sections within user-controlled files.
Capability inventory: The skill has the ability to write to the local filesystem and execute deployment commands via trusted CLI tools.
Sanitization: While the skill manages documentation merging, it relies on structured templates to isolate generated instructions from user-provided content.
[DATA_EXFILTRATION]: The skill accesses local configuration files to assist with development environment setup.
Evidence: references/makefile-macbook.md reads the user's ~/.ssh/config to resolve SSH aliases for remote MacBook deployment targets.

init-makefiles