A comparative study tested three context compaction strategies—FutureOS, OpenCode, and Codex—on their ability to retain information from agent sessions. FutureOS retained 83% of queryable information, significantly outperforming OpenCode (47%) and Codex (38%), with the key difference being that FutureOS preserves assistant prose while others compress it away. The analysis reveals that tool output dominates context volume but is rarely referenced, while the sparse assistant text is the primary source of follow-up questions.
codex-stats is a local analytics tool that aggregates session data from multiple coding assistants (Codex, OpenCode, Claude Code, Hermes) installed on a machine. It provides a browser dashboard with usage metrics, cost tracking, token analysis, and export capabilities across different time windows, with no API proxying or data interception.
Union Alpha, revealed as Pareto 26.9 from The Unbiased Co., is a stealth-launched model available on OpenRouter, OpenCode, and Cloudflare that uses a multi-model system architecture rather than single weights. The model experienced extreme demand causing performance issues, prompting AWS capacity expansion and introduction of paid access at competitive pricing starting at $2.50 per million input tokens.
Union Alpha, a stealth model with anonymous operators, launched free on OpenCode and OrcaRouter in September 2026. Community analysis suggests GLM-5.4 compatibility based on tokenizer fingerprinting, with benchmark performance near GPT-6 Astra levels at flash-tier pricing, though developer identity and exact specifications remain unconfirmed.
An Agent Skill enables coding agents to communicate like William Shakespeare, using a corpus of 14 plays and sonnets plus a portable Go binary for searching them. Compatible with opencode, Claude Code, Gemini CLI, and other Agent Skills clients following the open standard.
Mouse is an open source harness for long-running coding agents built on OpenCode. It passed 25 of 30 tasks on FrontierHarness Eval using Kimi K3, enforcing completion loops with verification rules to improve task completion accuracy.
Mouse coding agent achieved 23 of 30 passes on FrontierHarness tasks using GLM-5.3-Flash at $0.29 per pass, performing comparably to other agents on the full GLM-5.3 model while using significantly cheaper inference. Mouse missed two tasks that Kimi K3 passed on DeepSWE benchmarks but matched K3's performance on other shared failures.