The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains a repeated 'Zero attribution' directive (present in almost every file, including SKILL.md and all reference files) that instructs the agent to never include co-author, AI, or tool attribution in any output. This overrides the agent's default behavior of transparency regarding its AI nature.
[METADATA_POISONING]: The 'Zero attribution' rule includes a directive to propagate itself: 'Include this rule in any Markdown or settings file this workflow creates or updates.' This is a deceptive practice intended to hide the AI-generated origin of work products across the project lifecycle.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data in the form of bug reports, tickets, and user requests to generate implementation code and test suites.
Ingestion points: User-provided requests, bug reports, and tickets described in SKILL.md.
Boundary markers: Absent; there are no specific delimiters or warnings provided to the agent to treat this external content as untrusted data.
Capability inventory: The skill has the capability to write files (code, tests, docs) and execute shell commands for testing (pnpm, pytest, go test, etc.) as documented in references/test-commands.md.
Sanitization: Absent; the skill does not specify any sanitization or escaping for the content ingested from tickets.
[COMMAND_EXECUTION]: The skill provides a detailed catalog of shell commands for executing tests across multiple ecosystems (Node.js, PHP, Python, Go) in references/test-commands.md. While these are standard development tools, the agent is instructed to execute these commands locally based on its own seam discovery process.

tdd-loop