The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill documents the ord server utility, which is designed to process and serve untrusted Bitcoin inscription content, including HTML and JavaScript. This creates a surface where an agent reading inscription data could be influenced by malicious instructions embedded in that data. \n
Ingestion points: Multiple endpoints described in references/features-api.md and references/features-inscriptions-recursion.md (e.g., /content/ and /r/sat/) ingest and return arbitrary content from the Bitcoin blockchain. \n
Boundary markers: The skill lacks explicit instructions for the agent to use boundary markers or delimiters when handling this external content, although references/features-server-security.md warns that the environment should be considered untrusted. \n
Capability inventory: Sensitive capabilities documented in references/features-wallet.md (e.g., ord wallet send, ord wallet inscribe) could be abused if an agent follows instructions found within an inscription. \n
Sanitization: The documentation indicates that ord server does not sanitize inscription content, recommending network/domain isolation instead. \n- [COMMAND_EXECUTION]: The skill provides instructions for executing commands that manage Bitcoin wallets and local indices, which are powerful operations. \n
Evidence: references/features-wallet.md includes commands for sending Bitcoin/inscriptions and creating wallets. references/best-practices-reindexing.md documents the use of rm to delete the local index database.

ordinals