The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes multiple external project files (such as proposal.md, features/*.feature, design.md, and tasks.md) to drive its implementation logic and code generation. These files are treated as trusted input, creating a surface for indirect prompt injection where malicious instructions embedded in specifications could hijack the agent's workflow.
Ingestion points: status.yaml, proposal.md, beat/features/*.feature, design.md, tasks.md, beat/config.yaml.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are present when reading these artifacts.
Capability inventory: The skill possesses the ability to create and modify source files, update feature files, and execute shell commands via test runners.
Sanitization: No sanitization, escaping, or validation of the ingested content is defined in the workflow.
[COMMAND_EXECUTION]: The skill executes testing commands, specifically npx cucumber-js, to verify implementation against requirements. This is a functional requirement of the skill's purpose as an implementation tool.
[SAFE]: The skill incorporates strong process guardrails, such as a mandatory "Hard-Gate" requiring the use of git worktrees for environment isolation and enforcing TDD discipline before writing implementation code.

apply