The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions require the agent to execute shell commands directly via curl, bun run, and gh api. This includes sensitive operations like signing Bitcoin messages via bun run signing/signing.ts and performing inscriptions through bun run aibtc-news-classifieds/aibtc-news-classifieds.ts.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a managing editor for news 'signals' submitted externally, creating a surface where malicious content could influence payouts and blockchain inscriptions.
Ingestion points: External news signals are ingested via the news_signals tool.
Boundary markers: Decision logic relies on behavioral guidelines (4-Question Approval Test) rather than technical delimiters.
Capability inventory: The agent can execute financial transfers via aibtc__sbtc_transfer and Bitcoin inscriptions.
Sanitization: Verification of numeric claims is performed against external sources like mempool.space and Coinbase with specific tolerance thresholds.
[DATA_EXFILTRATION]: The skill transmits authentication data (signatures and timestamps) to https://aibtc.news using curl. While this domain belongs to the skill vendor (aibtcdev), the use of manual network requests for sensitive data is documented.
[EXTERNAL_DOWNLOADS]: The skill retrieves data from well-known services mempool.space and Coinbase for content verification.

aibtc-news-publisher