The Agent Skills Directory

[COMMAND_EXECUTION]: The instructions in SKILL.md direct the agent to build and execute shell commands using the RunCommand tool by interpolating user-provided values for file paths, document titles, and credentials. This pattern is vulnerable to command injection if a user or a malicious file content influences these variables to include shell metacharacters (e.g., using a title like \"; rm -rf /; #\").
[CREDENTIALS_UNSAFE]: The skill requires the agent to collect the sensitive FEISHU_APP_SECRET from the user and subsequently pass it as a plain-text environment variable in a shell command string. This method of credential handling is risky as the secret may be exposed in shell history, process logs, or through command injection vulnerabilities.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external Markdown files which may contain untrusted data.
Ingestion points: The scripts/write-to-feishu.js script reads file content from local paths using fs.readFileSync based on agent/user input.
Boundary markers: There are no delimiters or specific instructions for the agent or the processing script to ignore or sanitize embedded instructions within the processed Markdown.
Capability inventory: The skill has access to the RunCommand tool (shell execution), file system read access, and network access via fetch to the Feishu API.
Sanitization: No sanitization or validation of the input content is performed before it is parsed and sent to the external API.

feishu-doc-writer