The Agent Skills Directory

[PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data which could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: The new_text parameter in mcp__lsp__simulate_edit and mcp__lsp__preview_edit (SKILL.md), as well as diagnostic messages returned by mcp__lsp__evaluate_session (references/patterns.md).
Boundary markers: Absent. The skill instructions do not specify any delimiters or safety warnings for the agent to ignore instructions embedded within the simulated code or resulting diagnostics.
Capability inventory: The agent has the ability to write files to the disk using the mcp__lsp__commit_session tool (SKILL.md).
Sanitization: Absent. There is no evidence of input validation or escaping for the code strings or diagnostic outputs.
[COMMAND_EXECUTION]: The skill includes tools that allow the agent to perform permanent modifications to the host file system.
The mcp__lsp__commit_session tool, when used with the apply: true parameter, writes in-memory changes to the workspace disk.
[EXTERNAL_DOWNLOADS]: The skill references external resources required for its operation.
Documentation in SKILL.md points to github.com/blackwell-systems/agent-lsp as the required backend server for the simulation tools.

lsp-simulate