The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute various CLI tools and Git operations to manage the development lifecycle.
The instructions explicitly direct the use of git worktree, git rebase, and git merge --ff-only commands.
It utilizes environment-specific tools including openspec, change-linter, and grill-me for change management and developer interaction.
[INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it ingests and acts upon data from external, potentially untrusted sources.
Ingestion points: External data enters the context through user-provided requirements (Step 1), third-party review conclusions (Step 3), and raw source code containing conflict markers during the rebase process (Step 5) as defined in SKILL.md.
Boundary markers: While the skill defines structured "Acceptance" and "Merge" templates to guide output, it lacks explicit delimiters or instructions to ignore embedded commands within the ingested content from the user or third-party sources.
Capability inventory: The agent is granted capabilities to modify files, execute Git commands, and resolve code conflicts, which could be exploited if malicious instructions are embedded in the ingested data (SKILL.md).
Sanitization: There is no evidence of sanitization, escaping, or strict validation of the requirements or review text before they are processed by the agent.

wwwww