The Agent Skills Directory

[CREDENTIALS_UNSAFE] (HIGH): A hardcoded Supabase Anon Key (sb_publishable_w0enBaYGJ1jx8w2FNwpj4g_qDSYc5Oq) is embedded in multiple files, including cli/src/utils/config.ts, SKILL.md, and references/config.md. Hardcoding credentials, even those labeled as 'publishable', is a poor security practice that exposes the backend project to unauthorized interactions.\n- [EXTERNAL_DOWNLOADS] (MEDIUM): The installation script scripts/install.sh executes npm install, which downloads numerous dependencies from the public npm registry. These dependencies lack integrity verification (e.g., subresource integrity or pinned hashes), creating a surface for supply chain attacks.\n- [COMMAND_EXECUTION] (SAFE): The skill utilizes standard Node.js CLI patterns for its core logic. It properly manages its own configuration environment in the user's home directory (~/.openclaw/) and applies restrictive file permissions (0600) to sensitive configuration files.\n- [PROMPT_INJECTION] (LOW): There is a risk of Indirect Prompt Injection through processed marketplace metadata.\n
Ingestion points: The skill fetches user-controlled strings (titles and descriptions) from the Supabase API via getMarketplaceListings and getArtworkDetails in cli/src/utils/api.ts.\n
Boundary markers: None detected; external strings are integrated directly into the agent's context without delimiters.\n
Capability inventory: The skill possesses powerful capabilities to execute blockchain transactions (minting, buying, listing) via the viem library, which requires the use of a private key.\n
Sanitization: No evidence of sanitization or escaping of external content before it is processed by the agent.

clawbazaar-skill