The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions and command reference utilize cmux send and cmux send-key to interact with shell environments and terminal surfaces. It also provides instructions for executing a local script ./scripts/reload.sh as part of development workflows for contributors.
[PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection by enabling the agent to ingest content from terminal surfaces via the cmux read-screen command.
Ingestion points: The cmux read-screen command (documented in references/commands.md) reads terminal content from a specified surface into the agent's context.
Boundary markers: There are no explicit boundary markers or instructions to ignore embedded commands within the prompt templates provided in the skill.
Capability inventory: The skill possesses significant capabilities, including cmux send (command execution), cmux new-pane (environment modification), and the ability to execute local scripts like ./scripts/reload.sh (as seen in SKILL.md).
Sanitization: No evidence of sanitization or validation of the ingested terminal screen content is present in the provided instruction files.

cmux-workspace