
Mercury Agent v1.2.3 introduces Mercury Code, an autonomous AI software developer designed to handle end-to-end repository engineering workflows.
CosmicStack Labs has launched Mercury Code within the v1.2.3 release of Mercury Agent, providing a senior developer-grade AI coding agent accessible via the `/code` command. Rather than relying on external tools or simple snippet generation, Mercury Code executes the complete software engineering loop—inspecting repository context, planning implementations, modifying files, executing CLI tools, and validating results. The update emphasizes execution reliability and transparency, featuring automated error recovery to prevent workflows from stopping on initial failures, alongside real-time visibility into plans, reasoning traces, file diffs, and verification steps.
While terminal-based coding agents are proliferating rapidly, Mercury Code focuses on the primary pain point of autonomous coding: brittle execution and lack of observability. By integrating error-recovery loops with real-time reasoning previews directly inside an existing agent framework, it shifts the interaction from a fragile black box toward an inspectable peer engineering workflow.
• Complete engineering lifecycle: Handles repository exploration, planning, multi-file editing, testing, and verification natively within the terminal.
• Resilient task execution: Automatically recovers and retries when encountering blockers instead of abruptly halting.
• Transparent UX: Exposes task progress, reasoning previews, and diffs to provide developers with continuous visibility into the agent's actions.
DISCOVERED
1h ago
2026-09-14
PUBLISHED
2h ago
2026-09-14
RELEVANCE
AUTHOR
mercury__agent
