The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is entirely dependent on executing the lx CLI to interact with the Lexiong platform. It provides complex command patterns, including the use of shell-like loops and variable interpolation (e.g., using jq to parse block IDs and iterating over them to execute updates).
[PROMPT_INJECTION]: The skill facilitates Indirect Prompt Injection (Category 8) by ingesting untrusted data from an external knowledge base that is subsequently processed by the agent.
Ingestion points: Content is read into the agent's context using tools such as lx block find, lx block get, and lx block export (as described in SKILL.md and references/block-advanced.md).
Boundary markers: The instructions do not specify the use of delimiters or specific system instructions to prevent the agent from obeying commands embedded within the retrieved document blocks.
Capability inventory: The skill provides highly capable modification tools including lx block update, lx block delete, and lx block replace-section which can be leveraged if the agent is misled by data-embedded instructions.
Sanitization: There is no evidence of sanitization or filtering of the content retrieved from the knowledge base before it is used to influence the agent's decisions.
[SAFE]: No obfuscation, multi-layer encoding, or hidden content was found in the provided files. The skill correctly identifies itself as a tool for a specific vendor service ('tencent-lexiang') and uses resources consistent with that identity.
[SAFE]: The skill includes explicit security boundaries, instructing the agent to switch to other specialized skills for operations like creating new pages or pushing changes to Git, which demonstrates a principle of least privilege in its design.

lx-block