verify-submission

What This Skill Does

The verify-submission skill empowers the OpenClaw AI agent to act as a task administrator on the OpenAnt platform. It provides the necessary tools to handle the full lifecycle of task management, from vetting potential applicants to final verification of deliverables. By utilizing the OpenAnt CLI, the agent can programmatically interact with task states, ensuring that worker selection and work approval are handled efficiently and transparently.

Installation

To install this skill, run the following command in your terminal: clawhub install openclaw/skills/skills/ant-1984/verify-submission

Use Cases

• Applicant Vetting: Automatically filter or review applications for a bounty or task based on provided criteria or historical performance.
• Work Quality Control: Review submitted proofs or files against project requirements, providing structured feedback for rejections or confirming successful completion.
• Automated Payouts: Upon verifying a submission, the skill facilitates the workflow that triggers the release of escrowed funds to the contributor, streamlining the compensation process.
• Task Lifecycle Management: Keep task statuses updated by transitioning them from 'Open' to 'In Progress' and finally to 'Completed' as part of the operational workflow.

Example Prompts

1. "Check for any new applications on task_abc123 and if the applicant has a good rating, please accept them."
2. "Review the latest submission for task_xyz789; if it includes the required documentation link, approve it and leave a thank you note."
3. "Reject the submission from user_123 on task_abc123 because they didn't follow the formatting guidelines provided in the brief."

Tips & Limitations

• Always verify authentication: Before running operations, ensure the agent has an active session using npx @openant-ai/cli@latest status --json.
• Structured Output: The CLI is designed to output JSON, which the agent uses to parse data effectively. Always ensure the --json flag is appended to prevent command failures or parsing errors.
• Review Limits: Be aware that rejecting a submission is only possible if the task permits revisions; otherwise, the worker may not be able to fix their mistakes. Always review the maxRevisions parameter of your task settings before being overly critical.
• Human-in-the-loop: While the agent can automate these processes, ensure you provide clear rubrics. If the criteria are subjective, consider asking the agent to present findings for your final approval before executing the verification command.