The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool and Python subprocess modules to interact with the feishu-cli binary.
The Bash tool is scoped to feishu-cli, jq, and python3 as defined in the skill's YAML frontmatter.
Python scripts like fetch_chat_history.py and style_assets.py use subprocess.run to execute CLI commands for data retrieval and card linting.
[INDIRECT_PROMPT_INJECTION]: The skill includes functionality to ingest and process untrusted data from Feishu chat sessions, which is a potential injection vector.
Ingestion points: The fetch_chat_history.py script and msg history command retrieve message content, rich text (post), and interactive cards from external chat participants.
Boundary markers: The fetching script renders chat data into a structured timeline.txt file or NDJSON, effectively separating message content from administrative metadata.
Capability inventory: The skill has permissions to read/write local files (via Read and Write tools) and execute commands related to messaging operations.
Sanitization: The skill implements data sanitization for event processing, such as filtering event_id values to prevent path traversal when writing files to disk (found in references/workflows/event/workflow.md). It also uses standard JSON parsing (json.loads) to handle message payloads.

feishu-cli-messaging