The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Automated scanners identified multiple malicious URLs within the openfort.xyz domain (e.g., https://openfort.xyz/docs/recipes/). The skill consistently directs users to this blacklisted domain for documentation and sample integrations, which is a strong indicator of a phishing or typosquatting vector.
[CREDENTIALS_UNSAFE]: The skill provides instructions for the agent to access and manage sensitive blockchain credentials. It explicitly mentions the storage path ~/.config/openfort/credentials and references sensitive environment variables such as OPENFORT_SECRET_KEY and OPENFORT_WALLET_SECRET. In the presence of blacklisted external links, this creates a high risk of credential harvesting or exfiltration.
[METADATA_POISONING]: The skill metadata contains inconsistent homepage information. While the frontmatter lists openfort.io as the homepage, the instructional content redirects the agent and user to the blacklisted openfort.xyz domain, which may be a deceptive attempt to establish trust for a malicious site.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a wrapper for CLI operations that process untrusted external data. Ingestion points: Command-line interaction payloads (--interactions), contract addresses (--address), and player IDs (--player). Boundary markers: None are present to prevent the agent from misinterpreting embedded instructions in transaction data. Capability inventory: Full control over wallet creation, signing, and fund transfers via the openfort CLI. Sanitization: The skill instructions do not describe any sanitization or validation steps for incoming transaction data.

openfort-cli