The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests external specifications and requirement documents to generate implementation plans, which creates a surface for indirect prompt injection where adversarial content in those documents could influence the agent's code generation or command execution.
Ingestion points: User-provided requirement files and existing project source code as described in the "Before Writing" section of SKILL.md.
Boundary markers: The skill does not define specific delimiters or instructions to ignore potentially malicious embedded commands in the source requirement data.
Capability inventory: The agent is instructed to create/modify files and execute shell commands (git, pytest) as part of the implementation plan tasks.
Sanitization: No explicit sanitization or validation of the input specifications is described in the planning process.
[DYNAMIC_EXECUTION]: The skill involves a workflow where the agent generates Python code and shell commands (for testing and version control) based on high-level requirements and then executes them.
Execution method: The agent generates Python test cases and implementation code, then runs them using pytest and manages changes with git commands.

writing-plans