Skip to content

docs: add v2.2.0 changelog #84

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
saurabhhhcodes wants to merge 1 commit into
base: main
Choose a base branch
from
Open

docs: add v2.2.0 changelog #84

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
41 changes: 41 additions & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,47 @@ All notable changes to tokenfirewall will be documented in this file.
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).

## [2.2.0] - Unreleased

### Added

#### Smart Model Selection
- Added a planned Smart Model Selection release track for routing requests by task type instead of only by failure, context size, or cost.
- Documented built-in task categories for code generation, code review, math reasoning, document analysis, creative writing, translation, simple chat, data extraction, Chinese language tasks, vision, and documentation.
- Added guidance for task-type based model selection, including Claude-oriented code tasks, o1-oriented reasoning tasks, Gemini-oriented long-document tasks, and cost-efficient mini-model routing for simple tasks.
- Added configuration guidance for smart routing with task classification, fallback model selection, optional analytics, and custom detector hooks.

#### New APIs
- `classifyTask(prompt, context?)` for manual task classification before routing.
- `overrideTaskType(taskType)` for forcing the next request to use a specific task category.
- `getTaskAnalytics(options?)` for reviewing task distribution, model usage, estimated savings, and routing accuracy.
- Header-based overrides for request-level control, including disabling smart routing or forcing a task type for a single request.

#### Documentation
- Added `SMART-MODEL-SELECTION.md` as the primary design and usage guide for the v2.2.0 Smart Model Selection work.
- Added migration guidance for applications already using `createModelRouter()` with fallback, context, or cost strategies.
- Added operational notes for monitoring task analytics, tuning confidence thresholds, and handling ambiguous prompts.

### Changed

- The v2.2.0 routing plan extends the existing opt-in router architecture instead of replacing budget tracking, provider adapters, or cross-provider fallback.
- Smart routing remains designed as an opt-in capability so existing `patchGlobalFetch()`, budget guard, and fallback-router integrations continue to work without code changes.

### Migration Guide

1. Keep existing budget protection and provider setup unchanged.
2. Register provider API keys for every provider that smart routing may select.
3. Enable smart routing only where task-type routing is desired.
4. Start with the built-in task categories before adding custom detectors.
5. Review analytics after rollout and tune task definitions or confidence thresholds based on real traffic.

### Breaking Changes

- No breaking changes are announced for v2.2.0 at this stage.
- Smart Model Selection is documented as an opt-in routing layer, so existing v2.1.0 integrations should remain backward compatible.

---

## [2.1.0] - 2026-05-27

### Added
Expand Down