The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The management scripts scripts/_lib.sh and scripts/rotate_spaces_key.sh output newly generated Spaces access and secret keys directly to the console. While intended to allow users to save the one-time secret, this behavior exposes credentials to the AI agent's output buffer and potential platform logs.
Evidence: echo "Access Key: $access_key" and echo "Secret Key: $secret_key" in scripts/_lib.sh (line 186-187) and scripts/rotate_spaces_key.sh (line 52-53).
[COMMAND_EXECUTION]: The documentation and troubleshooting guides recommend using sudo for administrative operations, including tool installation and system clock synchronization.
Evidence: sudo ./aws/install in reference/aws-cli-operations.md and sudo ntpdate pool.ntp.org in reference/troubleshooting.md.
[PROMPT_INJECTION]: The script scripts/view_access_logs.sh downloads and searches through S3-compatible access logs. This creates an indirect prompt injection surface where malicious data in the logs could potentially influence the agent's behavior during log analysis.
Ingestion points: scripts/view_access_logs.sh (reading from bucket via aws s3 sync at line 104).
Boundary markers: Absent; log data is processed without delimiters or instructions to ignore embedded content.
Capability inventory: Shell command execution (grep at line 108), network operations (aws s3 sync).
Sanitization: No sanitization or content validation is performed on the downloaded log files before processing.
[EXTERNAL_DOWNLOADS]: The skill facilitates downloading the AWS CLI and syncing data from DigitalOcean Spaces endpoints. These operations target well-known and trusted infrastructure.
Evidence: curl "https://awscli.amazonaws.com/awscli-exe-linux-x86_64.zip" in reference/aws-cli-operations.md.

spaces