context-compressor
Public
RELEASE_VALIDATION.md
RELEASE_VALIDATION.md
npm run typecheck) — no cross-platform shim required this time.npm run typecheck:tests).npm test) — unlike 0.6.0's validation, which had to fall back to a Linux-hosted TypeScript-compiled assertion shim because the uploaded node_modules was Windows-built and unrunnable on that host. This build environment is Windows itself, so the canonical suite ran directly.npm run bench:static), model-free structural/determinism checks.lms dev installation in LM Studio desktop.npm run typecheck) — no cross-platform shim required this time.npm run typecheck:tests).npm test) — unlike 0.6.0's validation, which had to fall back to a Linux-hosted TypeScript-compiled assertion shim because the uploaded node_modules was Windows-built and unrunnable on that host. This build environment is Windows itself, so the canonical suite ran directly.npm run bench:static), model-free structural/determinism checks.lms dev installation in LM Studio desktop.qwen/qwen3.8-27b — 13/14 recall (matches the 2026-08-20 baseline; required 3 live fixes to get there: the truncation-headroom retry, the structural-digest fallback, and the compact rescue prompt — see CHANGELOG).deepseek-r1-distill-llama-8b — 9/14 recall.gemma-3-12b — 11/14 recall.meta/llama-3.2-3b — 13/14 recall (full prompt) vs 10/14 (compact prompt); this result is the basis for auto mode always using full.llama-3.2-8x3b-MoE — 14/14 recall (full prompt).qwen3-1.7b — 11/14 recall, sub-second chunks, /no_think honored.openai/gpt-oss-20b — 11/14 recall, no Harmony leaks.glm-4.7-flash — 9/14 recall.gemma-3-12b, google/gemma-4-26b, and mythomax-l2-13b all accept the system role on their LM Studio templates, so the no-system-role fold path stays dormant on these models — validated in the safe direction (the probe correctly declines to fold when folding isn't needed).qwen/qwen3.8-27b — 13/14 recall (matches the 2026-08-20 baseline; required 3 live fixes to get there: the truncation-headroom retry, the structural-digest fallback, and the compact rescue prompt — see CHANGELOG).deepseek-r1-distill-llama-8b — 9/14 recall.gemma-3-12b — 11/14 recall.meta/llama-3.2-3b — 13/14 recall (full prompt) vs 10/14 (compact prompt); this result is the basis for auto mode always using full.llama-3.2-8x3b-MoE — 14/14 recall (full prompt).qwen3-1.7b — 11/14 recall, sub-second chunks, /no_think honored.openai/gpt-oss-20b — 11/14 recall, no Harmony leaks.glm-4.7-flash — 9/14 recall.gemma-3-12b, google/gemma-4-26b, and mythomax-l2-13b all accept the system role on their LM Studio templates, so the no-system-role fold path stays dormant on these models — validated in the safe direction (the probe correctly declines to fold when folding isn't needed).