The Agent Skills Directory

[COMMAND_EXECUTION]: The skill generates and executes various shell scripts (dev-setup.sh, deploy.sh), Makefiles, and npm scripts. These scripts perform operations such as file manipulation (rm -rf, cp), package installation (npm install), and container management (docker, docker-compose).
[REMOTE_CODE_EXECUTION]: The deploy.sh script facilitates remote command execution by using ssh to connect to production and staging servers and run a sequence of commands (git pull, npm install, npm run build, pm2 restart).
[EXTERNAL_DOWNLOADS]: The skill's workflows rely on fetching external code and dependencies. It uses npm install and npm ci for package management and references official GitHub Actions (actions/checkout, actions/setup-node) and well-known third-party actions (codecov/codecov-action).
[INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection by processing untrusted input that flows into command execution.
Ingestion points: The deploy.sh script accepts environment names as command-line arguments ($1), and dev-setup.sh captures interactive user input ($REPLY) to determine execution flow.
Capability inventory: The skill possesses extensive capabilities including file system access (Read/Write/Edit), network operations (SSH, npm), and subprocess execution (Bash).
Boundary markers: No specific delimiters or instructions are provided to the agent to ignore potentially malicious content embedded in the data it processes.
Sanitization: The scripts do not implement sanitization or validation for the input arguments or captured user responses before using them in shell operations.

workflow-automation