The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/tui_explorer executes the Rust cargo build tool to compile and run local source code. This is a standard pattern for developer-oriented tools provided in source form.
Evidence: exec cargo run -q --manifest-path "$ROOT/cli/Cargo.toml" -p xtask -- tui-explorer "$@" in scripts/tui_explorer.
[COMMAND_EXECUTION]: The skill executes arbitrary terminal commands provided by the user via the --command flag to launch the target TUI application for exploration.
Evidence: skills/tui-explorer/scripts/tui_explorer discover --command "<app command>" as described in SKILL.md.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it processes and records untrusted output from external terminal applications.
Ingestion points: Terminal output, cursor positions, and screenshots from the application being explored (noted in references/discovery.md).
Boundary markers: The skill does not explicitly use boundary markers when interpolating discovered paths into the acceptance.md artifact.
Capability inventory: The skill can execute subprocesses via cargo and the target app command, and it performs file system write operations to the .agent-tui/ directory.
Sanitization: The skill implements basic normalization by stripping ANSI escape sequences and collapsing whitespace.

tui-explorer