The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides instructions for building applications that ingest and process untrusted terminal data, creating a potential surface for indirect prompt injection.\n
Ingestion points: Terminal message types such as tea.KeyPressMsg, tea.PasteMsg, and tea.MouseMsg are documented for use in SKILL.md and references/input.md.\n
Boundary markers: The examples do not include delimiters or specific instructions to ignore embedded commands in input data.\n
Capability inventory: The library documented by the skill supports powerful capabilities such as process execution (tea.ExecProcess), file writing (LogToFile), and network requests, as detailed in references/api.md and references/async.md.\n
Sanitization: No sanitization or escaping logic is included in the instructional code snippets.\n- [COMMAND_EXECUTION]: The skill documentation includes patterns for executing external system commands and processes.\n
Evidence: references/async.md contains an example using tea.ExecProcess and os.Getenv(\"EDITOR\") to hand over the terminal to a system editor like vi or code.

charmed-bubbletea