The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions and research strategies specify the use of the Bash tool to execute shell commands, primarily for git history analysis (git log, git diff) and searching the codebase (references/research-strategies.md).
[EXTERNAL_DOWNLOADS]: The skill is designed to perform research using WebFetch and WebSearch tools to retrieve documentation and best practices from external websites (SKILL.md, references/research-strategies.md).
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection due to its core workflow of ingesting untrusted data and using it to influence agent behavior. -- Ingestion points: Data is brought into the agent context via WebFetch, WebSearch, and by reading files from the local filesystem (SKILL.md, references/research-strategies.md). -- Boundary markers: The provided instructions do not specify the use of delimiters or 'ignore' instructions to separate untrusted content from the system prompt. -- Capability inventory: The skill has access to powerful tools including Bash, Task/Explore, and the ability to modify files as part of its execution phase (SKILL.md, references/examples.md). -- Sanitization: There is no mention of sanitizing, escaping, or validating external content before it is processed by the LLM to generate questions or an execution plan.

prompt-improver