The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill reads and acts upon instructions provided in several repository files, creating a significant surface for indirect prompt injection. If these files are modified by an untrusted party (e.g., via a malicious Pull Request), they could influence the agent to generate malicious infrastructure code or execute dangerous commands.
Ingestion points: The skill reads specs/TASKS.md, specs/ARCHITECTURE.md, specs/DATABASE.md, specs/SECURITY.md, specs/CODE_STYLE.md, and specs/TESTING.md to determine its actions and the content of its outputs.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to disregard natural language instructions that might be embedded within these configuration and spec files.
Capability inventory: The skill has the capability to write files (e.g., Dockerfile, .github/workflows/ci.yml), execute shell commands (e.g., docker build, linting, testing), and perform git commits.
Sanitization: The instructions do not specify any validation or sanitization of the content read from the spec files before it is used in file generation or command execution.
[DYNAMIC_EXECUTION]: The skill dynamically generates executable configuration files, such as Dockerfile and GitHub Actions workflows, based on templates and project-specific metadata. It also executes verification commands (lint, test, build) that are derived from the project's specification files.
The skill generates Dockerfile stages and entrypoints based on the detected stack.
The skill constructs .github/workflows/ci.yml using standard actions for Node.js and Python.
The skill attempts to run local verification commands like docker build and test suites defined in specs/TESTING.md.

devops-infrastructure-engineer