◉CLAUDE LABJP
●2.1.296 — Subagents can set their own autoCompactWindow, and an env var pins the model used by workflow agents●AGENTS — Managed Agents gain dynamic workflows in beta: an agent can build a run for work with many pieces (Oct 9)●SONNET 4.5 — Retires on the Claude API on 11/30, 50 days left. Migrate to Sonnet 5.5●Q&A — A long-running report: Max plan users hit usage limits almost immediately●HAIKU 5.5 — Code written for Haiku 4.5 hits a 400 error on budget_tokens●NEW — Before you cut an always-loaded instruction file in half, sort it into three drawers●2.1.296 — Subagents can set their own autoCompactWindow, and an env var pins the model used by workflow agents●AGENTS — Managed Agents gain dynamic workflows in beta: an agent can build a run for work with many pieces (Oct 9)●SONNET 4.5 — Retires on the Claude API on 11/30, 50 days left. Migrate to Sonnet 5.5●Q&A — A long-running report: Max plan users hit usage limits almost immediately●HAIKU 5.5 — Code written for Haiku 4.5 hits a 400 error on budget_tokens●NEW — Before you cut an always-loaded instruction file in half, sort it into three drawers
TAG

retrieval augmented generation

1 articles
← Back to all tags
Related:
Claude API1RAG1evaluation1TypeScript1
⬡ API & SDK/2026-07-09Advanced

When the RAG Started Being Confidently Wrong — Field Notes on Measuring Retrieval Misses With Groundedness

In a Claude API RAG, the answers stay fluent while the facts drift. Often the cause is a silent recall decay on the retrieval side, missing the document that holds the answer. Field notes on measuring groundedness and retrieval hit rate and walking the system back, with working code and real numbers.