The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Python subprocess module across its script suite to execute local blockchain binaries and scripts from user-provided directories.
In scripts/run_basic_network.py and scripts/run_mixed_network.py, it spawns validator processes such as validator-engine to establish the local test network.
In scripts/compile_tolk.py, it invokes the tolk compiler binary to transform source code into blockchain-compatible artifacts.
In scripts/run_fift_script.py, scripts/wallet_send.py, and scripts/build_stateinit.py, it executes the fift interpreter to process contract assembly and sign messages.
In scripts/ton_triage_lib.py, it uses sys.executable to run generate_tl.py from the local repository for API binding generation.
[PROMPT_INJECTION]: The skill processes data from blockchain states and validator logs, which constitutes an indirect prompt injection surface.
In scripts/inspect_latest_transaction.py, it fetches and displays transaction comments and message bodies that may contain untrusted content from the blockchain.
In scripts/summarize_run.py, it parses validator and error logs for debugging markers and user-defined patterns, which involves reading content potentially influenced by external events.

ton-bug-triage