The Agent Skills Directory

[COMMAND_EXECUTION]: The skill automatically identifies and executes commands defined in the repository's configuration files (such as package.json scripts, Makefile, build.gradle, or pom.xml) to run tests and linters. This provides a mechanism for the execution of arbitrary shell commands as part of the TDD workflow.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the ingestion of external data which may contain malicious instructions designed to influence the orchestrator or its subagents.
Ingestion points: External requirement documents provided via URLs, local plan files (docs/plans/*.md), and project-specific instruction files like CLAUDE.md or AGENTS.md (as described in the 'Specs & Commands Discovery' section).
Boundary markers: The skill attempts to maintain boundaries by providing subagents with specific task headings and authoritative plan paths, but it does not specify rigorous sanitization or escaping of the content fetched from external URLs.
Capability inventory: The orchestrator dispatches subagents with file-write permissions and executes shell commands discovered in the codebase.
Sanitization: There is no evidence of content sanitization for data processed from external requirement links or wiki pages before it is used to drive the implementation plan.
[EXTERNAL_DOWNLOADS]: The skill allows users to provide a link to an external requirement or design document (e.g., a wiki page or ticket URL). This remote content is fetched and processed to determine the agent's implementation strategy.

ss-coding