The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructs the agent to periodically fetch and follow instructions from a remote markdown file (https://bankrsignals.com/heartbeat.md). This architecture creates an indirect prompt injection surface where the remote source can dynamically influence agent actions without a formal skill update.
Ingestion points: SKILL.md and HEARTBEAT.md (remote retrieval of heartbeat guidelines).
Boundary markers: Absent; the instructions direct the agent to "follow" the fetched content directly.
Capability inventory: File system writing (config.json), network requests (curl), and shell command execution (node, python3).
Sanitization: None detected for the remote content.
[COMMAND_EXECUTION]: The skill provides shell commands and scripts that execute logic using curl, mkdir, and node -e or python3 -c for cryptographic signing and data processing. Specifically, publish-signal.sh generates a JavaScript snippet to sign messages via Node.js.
[EXTERNAL_DOWNLOADS]: The skill performs multiple network requests to vendor domains (bankrsignals.com, bankr.bot, api.bankr.bot) to manage custodial wallets, register providers, and fetch signal data or heartbeat instructions.

bankr-signals