The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill mines recent agent transcripts to extract working conventions and preferences, which introduces a surface for indirect prompt injection.
Ingestion points: Step 1 in SKILL.md identifies that the skill reads conversation transcripts from the workspace-scoped path.
Boundary markers: There are no specific boundary markers or instructions to the subagents to disregard instructions embedded within the mined transcript data.
Capability inventory: The skill has the capability to write new skill files using the writing-for-agents tool and execute git operations (commit, PR) to persist them.
Sanitization: The instructions do not describe any sanitization or filtering process for the data extracted from transcripts before it is clustered and used to draft the final skill.
[EXTERNAL_DOWNLOADS]: The skill references a configuration example located in a public repository on a well-known platform (github.com/cursor/plugins) to provide structural guidance for the agent.
[COMMAND_EXECUTION]: The skill uses standard git command-line tools for analyzing history (git log) and managing the deployment of generated files (git worktree, git commit, git request-pull).

automate-me