Apiyi — Claude Code v2.1.92 MCP Result Persistence
Public MindPattern findings, entities, and graph evidence that cite this source.
Findings
2
All-time hits
2
High value
1
Last seen
2026-04-08
Connected entities
Related findings
- 2026-04-08 / SKILLSClaude Code MCP 500K: Tool Result Persistence Override Lifts MCP Response Limit from Default to 500,000 CharactersMCP servers can now annotate tool results with _meta['anthropic/maxResultSizeChars'] to pass through up to 500K characters, eliminating the previous truncation bottleneck for large database schemas, log outputs, and file reads. This is a per-result override, not a global setting, allowing MCP servers to selectively expand limits for results that need it while keeping defaults tight for everything else.
- 2026-03-28 / TOOLSGLM-5.1 Launches: 744B MoE Scores 94% of Opus on Coding, Open Weights April 6-7Z.ai (formerly Zhipu AI) launched GLM-5.1 on March 27, a 744B parameter MoE model (256 experts, 8 active per token, 40B active params). Self-reported coding evaluation score of 45.3 vs Opus 4.6's 47.9 (94.6% parity), a 28% improvement over GLM-5's 35.4. Open weights confirmed for April 6-7 release. Important caveat: benchmarks are self-reported with no independent verification. At these scores and with open weights, GLM-5.1 could become the strongest open-weight coding model available.