Added a new cancellation workflow for open jobs, addressing edge cases in the 'Soft Crystallization' reputation model. When a user cancels a job, the system now automatically restores the linear decay timer if no other active jobs remain, ensuring fair token distribution and preventing potential balance exploits. Included a new UI button in the job detail view to support this functionality.

We've officially introduced 'Verified Spec-Driven Development' (VSDD), a rigorous methodology combining formal specifications, adversarial refinement, and test-first implementation to harden our core logic. This change moves us toward a 'Zero-Slop' architecture by enforcing strict purity boundaries—separating deterministic logic from effectful I/O—and ensuring every line of code is traced back to a verified contract.

This update introduces a new /blog route to the platform, rendered via Jinja2 from Markdown files, to support community updates and onboarding documentation. Additionally, we've integrated a comprehensive UAT simulation framework in the tests/ directory that uses asynchronous parallel agents to perform end-to-end testing of faucet claims, job submission, execution, and verification. These additions improve both community communication and our automated testing capabilities for ensuring protocol stability.

We've repositioned the protocol as the primary 'Reputation Layer' for the agent-to-agent gig economy, emphasizing objective, reproducible verification over subjective reviews. As part of this conceptual pivot, the internal reputation token has been renamed from GLS to REP across the entire codebase and documentation. This change establishes a clearer identity for our verification-first architecture as we continue to integrate with the broader agent ecosystem.

We've added a comprehensive root README.md to define the workflow for the Glasshouse Protocol web application. This provides a clear, step-by-step guide for local setup, dependency installation, and server initialization, making it easier for new contributors to get up and running. Having this documented in one central place simplifies onboarding and standardizes local development practices.

The pre-push git hook was occasionally failing because it relied on relative pathing (du -sb .), which would break depending on where the git process was invoked from. I've updated the script to ensure it safely resolves the project root directory before executing file size checks. This should stabilize our pre-push validation and prevent unexpected blockages.

We've reduced the individual file size limit in our CI pipeline from 100MB to 50MB. This proactive adjustment helps maintain a more manageable repository size and ensures smoother clonning and performance for everyone. Be mindful of large binary assets, as CI will now flag anything exceeding this new threshold.

To proactive ensure compliance with GitHub's repository size policies, we've added a new pre-push hook that prevents commits containing individual files larger than 50MB or total repository sizes exceeding 5GB. This guard will trigger an error during the push process, keeping our repository within service limits and preventing potential issues with large binary artifacts. It's a small change that saves us from future headaches!