The Agent Skills Directory

[PROMPT_INJECTION]: The skill analyzes content from external repositories (Dockerfiles, compose files, and documentation) to generate deployment configurations. This introduces an indirect prompt injection surface where a malicious repository could potentially influence the agent's logic for generating stack entries.
Ingestion points: Analysis of source repository compose files, Dockerfiles, and deployment documentation in Step 1 of SKILL.md.
Boundary markers: No specific boundary markers or instructions to disregard embedded commands in the source files are present.
Capability inventory: The skill generates compose.yml, .env templates, and mise.toml tasks which involve shell command execution (docker compose pull, docker compose up).
Sanitization: No explicit sanitization or validation of the ingested external content is mentioned.
[COMMAND_EXECUTION]: The skill generates shell command snippets for mise.toml tasks. These commands are intended to automate local orchestration of Docker deployments and represent the primary functional purpose of the tool.
[DATA_EXFILTRATION]: The skill facilitates the creation and management of .env files, which may contain sensitive credentials. It follows security best practices by using .env.example as a template and storing runtime data in specific user-controlled paths (/home/delorenj/data/), with no evidence of unauthorized network transmission of this data.

stacks-deploy