The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies heavily on the execution of the lark-cli binary and system commands like date to perform its functions.
The workflow executes lark-cli vc +search, lark-cli vc +notes, and lark-cli drive metas batch_query to retrieve data.
It also uses lark-cli docs +create and lark-cli docs +update to write reports back to the cloud environment.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and summarizes untrusted data from external meeting records.
Ingestion points: Meeting titles and notes are fetched via lark-cli vc +search and lark-cli vc +notes in SKILL.md.
Boundary markers: There are no delimiters or explicit instructions to ignore embedded commands within the processed meeting content.
Capability inventory: The agent has the ability to write to files and create documents via lark-cli docs +create and lark-cli docs +update (as referenced in SKILL.md and its dependencies).
Sanitization: No evidence of sanitization or escaping of the fetched meeting content is present before it is interpolated into the summary report or document creation commands.

lark-workflow-meeting-summary