The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documents commands that fetch and install software from remote sources, including CLI updates via bl update (reference/update.md), Command Packs via bl plugin install (reference/plugin.md), and agent skills via bl skill add (reference/skill.md).
[COMMAND_EXECUTION]: The skill operates by executing the bl command-line utility. This includes sensitive operations documented in reference/auth.md for managing API keys and Access Key Secrets, and file upload capabilities in reference/file.md and reference/knowledge.md.
[DYNAMIC_EXECUTION]: Functional logic can be executed dynamically through bl pipeline run (reference/pipeline.md), which parses and runs JSON/YAML workflow definitions, and bl mcp call (reference/mcp.md), which executes tools via the Model Context Protocol.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data that could contain malicious instructions. Ingestion points: Input is accepted via --prompt, --message, and --query flags in app, chat, and search commands. Boundary markers: No delimiters are specified to isolate untrusted content. Capability inventory: The skill has the power to perform file uploads and install plugins. Sanitization: No input filtering or validation is documented for prompt-interpolated values.

bailian-cli