# Changelog

## v0.12.1

- Added a follow-up patch release on top of `v0.12.0` with more detailed release documentation for the new `MHC Atlas OS` monorepo flow
- Refreshed committed demo history snapshots under the bundled demo projects so shipped examples reflect the latest pipeline runs
- Kept the current modular API, UI, parsing, comparison, prioritization, policy, reporting, and decision-persistence stack unchanged from `v0.12.0`
- Preserved the current conservative framing: interpretable structure-guided decision support, not affinity prediction, immunogenicity prediction, or proof of mechanism

## v0.12.0

- Added the `MHC Atlas OS` monorepo scaffold with dedicated `apps`, `agents`, `biology`, `core`, `storage`, and `tests` layers
- Added Biopython-based structure parsing for PDB and mmCIF files with residue, chain, coordinate, and confidence summaries
- Added WT-vs-mutant structural comparison with residue changes, CA shift metrics, and confidence delta tracking
- Added an interpretable prioritization engine with policy-based post-processing and human-readable explanations
- Added a FastAPI backend with `/parse`, `/compare`, `/rank`, and `/pipeline` endpoints
- Added a Streamlit UI for parsing, comparison, ranking, interpretation, confidence review, and impact summaries
- Added decision report generation and SQLite-backed decision history persistence
- Added thin workflow agents, demo structures, seed scripts, and deterministic golden-path tests

## v0.10.5

- Added a canonical `golden_weekly_review_demo` as the primary evaluator workflow
- Added stronger first-run, pilot-workflow, and release-facing onboarding docs
- Improved app orientation for project and workspace mode
- Polished weekly review packets, decision packets, and role-oriented views for demo clarity
- Added buyer-safe framing around target users, workflow value, and conservative scope

## v0.10.0

- Added multi-project workspace configs and inventory exports
- Added project history snapshots and change summaries
- Added weekly review packets and decision packets
- Added scientist, computational lead, and manager role views
- Added next-action and open-question generation
- Added workspace-aware app and CLI flows for recurring review cycles

## v0.9.0

- Added pilot workflow support with review queues, shortlists, and next-action artifacts
- Added structured feedback capture, summaries, and feedback trend outputs
- Added local session logging, action summaries, and pilot usage analytics
- Added checklist templates and checklist-run exports
- Added collaborator handoff bundle creation with explicit conservative scope messaging
- Extended the Streamlit app and CLI for review, feedback, handoff, and checklist flows

## v0.8.0

- Added Python packaging via `pyproject.toml`
- Added unified `mhc-atlas` CLI
- Added install docs, quickstart docs, and demo docs
- Added Docker and devcontainer support
- Added environment checks and demo bundle validation

## v0.7.0

- Added interactive local app
- Added scenario analysis and comparison
- Added evidence drilldown and demo projects