Rare find

Lynkr. A middleman between your code and AI models that cuts down what you pay to use them.

fast-editor.github.io/Lynkr

Vaya's read on this project

Problem, audience, market, and the verdict — sign in to see it.

Updates

July 2026
  • Added codex wrap support
  • Removed comments
  • Added tiers
  • chore: ignore local handoff notes
  • feat(routing): WS6 cascade verification — verify cheap-tier answers, …
  • chore: untrack local-only files (plan ledger, litellm config); fix do…
  • chore: untrack marketing folder (local-only drafts)
  • docs: restore 53% tool-token-reduction claim
  • docs: routing-intelligence + benchmarking guides; README numbers from…
  • bench: hermetic across runs + similarity-aware cache assertion
  • bench: 8 routing-regression scenarios + correctness scoreboard
  • feat(routing): propensity logging, learning loop, fingerprint session…
  • feat: add Eden AI provider (OpenAI-compatible, 600+ models, EU/GDPR) …
  • Added telemetry
  • fix(agents): make the sub-agent learning subsystem actually work (#80)
June 2026
  • fix(init): correct Bedrock + LM Studio wizard schema
  • feat: Multi-Tool Wrap + OAuth Subscription Routing + Headroom Integra…
  • Release —v9.6.0 — Routing resilience & output quality
  • feat(routing): tier-aware fallback, output format guard, task decompo…
  • fix(prompt-cache): repair Bedrock Converse mapping and gate injection…
  • fix(responses): repair blank/orphan tool_call_id before provider disp…
December 2025
  • Release —Windows Release 0.1.0