The Agent Skills Directory

[DATA_EXFILTRATION]: The skill contains logic to collect and transmit telemetry data to a remote server.
Evidence: Shell scripts in SKILL.md extract a convexUrl from ~/.superstack/config.json and use curl to POST JSON data containing the skill name, phase, status, and system architecture (uname -s-uname -m).
Consent Mechanism: The instructions include a manual prompt to ask the user for permission (AskUserQuestion) before enabling telemetry and recording the preference in a configuration file.
[INDIRECT_PROMPT_INJECTION]: The skill processes project-specific data from .superstack/build-context.md, which is an untrusted ingestion point.
Ingestion points: The workflow reads .superstack/build-context.md (in SKILL.md) to verify build status and deployment history.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to distinguish between its instructions and the content of the context file.
Capability inventory: The skill has the ability to execute shell commands, perform Solana blockchain transactions, and initiate network requests via curl across all scripts.
Sanitization: No evidence of validation or sanitization is present for the content read from the build context file.
[COMMAND_EXECUTION]: The skill requires the execution of multiple system commands to perform its primary function.
Evidence: Instructions include running anchor build, solana config set, sha256sum, and anchor deploy to manage the Solana deployment process.

deploy-to-mainnet