The Agent Skills Directory

[NO_CODE]: The skill package is comprised entirely of Markdown documentation files and does not include any executable scripts, binaries, or system automation code.\n- [INDIRECT_PROMPT_INJECTION]: The skill uses local context files to personalize its advice, creating a surface where untrusted data could influence agent behavior.\n
Ingestion points: The instructions in SKILL.md direct the agent to read profile and context data from .agents/em-context.md and report-specific files in the .agents/reports/ directory.\n
Boundary markers: There are no explicit delimiters or protective instructions to ensure the agent treats the ingested data strictly as information rather than commands.\n
Capability inventory: The skill uses file system read and write operations for managing context and provides natural language responses based on the collected data.\n
Sanitization: No technical sanitization or validation logic is applied to the content stored in or read from the context files.\n- [EXTERNAL_DOWNLOADS]: The skill includes links to external resources for management guidance and source materials.\n
Details: Educational links to manager.dev, beehiiv.com, terriblesoftware.org, and refactoring.fm are documented in references/sources.md as references for the user. These do not involve automated network fetching by the skill.

management-transitions