The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill generates and manages a sensitive Veil private key (VEIL_KEY) stored in ~/.clawdbot/skills/veil/.env.veil. It also reads an API key from ~/.clawdbot/skills/bankr/config.json to authenticate requests to the Bankr API. Access to these credentials is central to the skill's functionality.
[EXTERNAL_DOWNLOADS]: The skill depends on the @veil-cash/sdk package, which is recommended to be installed via NPM or cloned from the veildotcash/veildotcash-sdk GitHub repository. These sources are outside the predefined trusted vendor list.
[COMMAND_EXECUTION]: Numerous bash scripts are used to invoke the veil CLI and handle blockchain data via system commands such as jq, curl, and node.
[DATA_EXFILTRATION]: User prompts and unsigned transaction payloads are transmitted to api.bankr.bot for signing and execution. This endpoint is a vendor-owned resource associated with the skill author.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface where untrusted data may be processed and forwarded to an external API.
Ingestion points: scripts/veil-bankr-prompt.sh takes arbitrary command-line arguments as input; scripts/veil-bankr-submit-tx.sh reads transaction JSON from an input stream.
Boundary markers: Absent. There are no delimiters or instructions to ignore embedded commands in the data sent to the Bankr API.
Capability inventory: The skill can perform network operations via curl and execute private blockchain transactions via veil_cli.
Sanitization: Data is encoded into JSON using jq, which prevents structural breakage but does not filter the content for malicious instructions.

veil