The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the agent to run several local Python scripts including prepare_context.py, fetch_diaries.py, and master_diary_sync.py, as well as external CLI tools like qmd embed. These scripts perform file system operations and network requests to external APIs.
[PROMPT_INJECTION]: The instructions in SKILL.md use "Highest Priority" and "Ironclad" rules to override the agent's default behavior, explicitly forbidding conversational output between steps and demanding an autonomous "one-shot" execution path. It also commands the agent to set SafeToAutoRun: true to bypass platform-level user confirmation for shell commands.
[INDIRECT_PROMPT_INJECTION]: The workflow ingests content from the user's project environment, including README.md and various diary files, to generate summaries and context. Maliciously crafted content within these project files could potentially influence the agent's logic during the fusion and synchronization phases.
[DATA_EXFILTRATION]: The skill is designed to transmit project summaries and diary logs to the Notion API. Due to the high degree of automation and the "one-shot" execution model, there is a risk that sensitive information captured from the local project environment could be exfiltrated to the cloud service without granular user review of the specific content being sent.

diary