The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent to execute local shell commands such as grep, find, and ls to analyze codebase coupling and locate existing tests (found in references/coupling.md and references/phase-5-test.md).
[COMMAND_EXECUTION]: In Phase 5, the agent is instructed to run the project's native test suite to verify that the implementation meets the stated goals.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because it processes external, untrusted task data to guide its actions.
Ingestion points: The agent reads task descriptions, design documents, and discussion histories from the project environment (references/phase-1-brainstorm.md).
Boundary markers: There are no explicit instructions to the agent to treat this ingested content as data rather than instructions, nor are there delimiters to prevent embedded instructions from influencing the agent's behavior.
Capability inventory: The agent has the ability to write to the filesystem and execute shell commands (searching and testing).
Sanitization: The skill does not provide mechanisms for validating or sanitizing the content of the ingested task data.

improve