The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructs the agent to treat solveInstructions from an external API response as the "authoritative challenge-specific instruction block." This creates a conflict with general safety guidelines and may lead the agent to follow malicious instructions if the external source is compromised.
[PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection attacks.
Ingestion points: Challenge payloads (including doc, questions, and solveInstructions) fetched via curl from coordinator.agentmoney.net.
Boundary markers: The skill includes a "Golden Rule" warning that coordinator response payloads are challenge data and not trusted system instructions, though this is partially contradicted by the specific instruction to treat them as authoritative.
Capability inventory: The skill uses curl for network communication and has the capability to perform financial transactions (swapping, staking, submitting) through the bankr skill.
Sanitization: No automated sanitization or filtering of the external instruction content is performed before processing.
[COMMAND_EXECUTION]: The skill relies on curl and jq to communicate with external APIs and process JSON responses, which involves handling the BANKR_API_KEY and transaction data.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to coordinator.agentmoney.net and api.bankr.bot. These are vendor-owned domains related to the skill's intended functionality.

botcoin-miner