The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill manages sensitive configuration and cryptographic material.
Evidence: References ~/.bitvm-bridge/bridge.toml for storing Bitcoin private keys (depositor, operator, verifier, withdrawer).
Evidence: Mentions usage of .env files for AWS credentials (BRIDGE_AWS_ACCESS_KEY_ID, BRIDGE_AWS_SECRET_ACCESS_KEY).
Evidence: CLI commands involve passing secret keys as plaintext arguments (e.g., bridge keys -d <SECRET_KEY>).
[EXTERNAL_DOWNLOADS]: Fetches data from external specialized services.
Evidence: Downloads Bitcoin header data from zerosync.org using wget.
[REMOTE_CODE_EXECUTION]: References external toolchain installation.
Evidence: Provides instructions to install the Risc0 toolchain via https://dev.risczero.com/api/zkvm/install.
[COMMAND_EXECUTION]: Requires building and running local binaries.
Evidence: Commands for cargo build and execution of generated binaries like bridge and prover for cryptographic operations.
[PROMPT_INJECTION]: The skill processes external data which could serve as a vector for indirect prompt injection.
Ingestion points: Bitcoin header files (prover/data/mainnet-headers.bin) and CLI arguments for transaction IDs and addresses.
Boundary markers: None identified in the provided documentation.
Capability inventory: Executes subprocesses (cargo build, local binaries), performs network operations via wget, and performs file system operations.
Sanitization: No explicit sanitization or validation of the ingested binary data or CLI inputs is described.

bitvm