Skip to content

CHANGELOG.md

Latest commit

 

History

History
192 lines (162 loc) · 11.4 KB

CHANGELOG.md

File metadata and controls

192 lines (162 loc) · 11.4 KB

Changelog

All notable changes to Forest will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

0.6.0 - 2026-02-13

Added

  • Temporal graph analysis toolkit (#34)
    • forest diff --since ... for node/edge change summaries across snapshots
    • forest growth for graph growth metrics over time
    • forest snapshot for explicit point-in-time captures
  • Snapshot persistence model with automatic daily snapshots plus manual snapshot API/CLI parity
    • New graph_snapshots table and temporal analysis core
    • New API routes: GET /api/v1/graph/diff, GET /api/v1/graph/growth, GET /api/v1/graph/snapshots, POST /api/v1/graph/snapshots
  • Node history and restore workflow (#31)
    • New node history capture and restore commands
    • Historical seed support for existing nodes when first accessed
  • Ranking and operations controls
    • forest admin rescore for full-graph rescoring
    • Degree counter maintenance and safeguards for stale edge metrics
  • Deployment ergonomics
    • deploy/update-remote.sh for repeatable server updates
    • Build fallback and startup health-check retry/backoff for safer deploys

Changed

  • Ranking quality improvements in edge scoring with stronger tag/project signal handling and improved rescore behavior
  • TypeScript script consistency by standardizing scripts on bun x tsc
  • Web node detail UX
    • Connected node list now sorts by strongest accepted edge first
    • Connected nodes are directly clickable for navigation
    • Node markdown renders as HTML by default with a source toggle
    • Content container now grows naturally with content (no fixed-height scroll box)

Fixed

  • Manual snapshot duplication - creating a manual snapshot no longer triggers a second automatic snapshot
  • Accepted-degree counter drift - rebuilt and hardened counter updates to keep graph stats accurate over time
  • CLI runtime compatibility - fixed tag flag normalization and ESM clerc loading edge cases
  • Deploy false negatives - startup health checks now tolerate slower service warm-up

0.3.1 - 2025-10-22

Added

Git-Style Node References & Progressive IDs

  • Progressive UUID abbreviation - IDs now display with minimum length needed for uniqueness (7e603d1, 039aa44)
    • Starts at 4 characters, grows as needed
    • Backward compatible: all existing 8-char and full UUIDs still work
    • formatNodeIdProgressive() replaces fixed 8-char display
  • Recency references - Git-style HEAD~N syntax for recent nodes (7e603d1)
    • @ or @0 - Most recently updated node
    • @1, @2, etc. - Second, third most recent
    • Works across all commands: forest node read @, forest node link @ @1
  • Tag search references - Find nodes by tag (7e603d1)
    • #typescript - Node tagged with 'typescript'
    • Must match unique node or shows disambiguation
  • Title search references - Find nodes by title substring (7e603d1)
    • Quoted strings match against node titles
    • Must be unique or shows Git-style disambiguation
  • Shell completions - Tab completion for bash and zsh (7e603d1)
    • completions/forest.bash and completions/forest.zsh
    • Supports commands, flags, and recency references

Modular Formatter Architecture

  • Centralized formatting system in src/cli/formatters/ (c01acf4, ac427de)
    • Reusable formatter functions for consistent output
    • Color scheme system with customization
    • Markdown formatting utilities
  • Applied to edge and stats commands for better visual hierarchy (10a09b7)
  • Comprehensive color styling across all CLI commands (8a68dc7)
    • Forest color scheme as default (dfb3b5a)
    • Configurable color schemes (2c0f51e)

Document Model Improvements

  • Canonical document model - First draft implementation (bf7dfd0)
    • Document session system for editing multi-chunk documents
    • src/cli/shared/document-session.ts with comprehensive tests
  • Preserve existing tags when saving document segments (a2e350b)
    • Tags are no longer lost during document edits
  • PR review improvements - Comprehensive refinements (ef66a25)
    • Enhanced error handling, validation, and edge cases
  • Explicit TypeScript type configuration for consistent resolution (3df37ae)

CLI Clarity

  • Clarified search vs. explore roles (7902c2f, 8543b8f)
    • search - Semantic search for specific content
    • explore - Interactive graph traversal and discovery
    • Better help text and documentation

Changed

  • Enhanced CLI visuals - General improvements to output formatting (7ad5372)
  • Documentation reorganization - Moved docs to docs/ directory (bd871ac)
    • Created 25+ task planning documents in docs/tasks/
    • Added comprehensive architecture guides

Removed

  • Cleanup of cruft - Removed ~547K lines of old output files (9bb252a, bd871ac)
    • Deleted old outputs/ directory with generated files
    • Removed TLDR doc generators (moved to standalone repo)
    • Removed obsolete export and test files

Fixed

  • TypeScript configuration - Resolved module resolution inconsistencies (3df37ae)
  • Tag preservation - Tags no longer lost during document segment saves (a2e350b)

0.3.0 - 2025-10-22

Added

REST API & WebSocket Events

  • Complete REST API with full CRUD endpoints for nodes, edges, and tags under /api/v1 (5f6170c)
    • Advanced edge management: accept, reject, explain, promote, sweep, undo
    • Pagination support with offset-based navigation
    • Consistent error handling and response envelopes
  • WebSocket endpoint (/ws) with real-time event streaming (5f6170c)
    • Event subscriptions with tag filtering
    • Broadcasts all graph mutations: node:created, node:updated, node:deleted, edge:*, tag:renamed
  • Elysia server foundation with health and stats endpoints (139f730)
  • Event bus system for broadcasting graph mutations (5f6170c)

TLDR Standard for Agent Discovery

  • TLDR Standard v0.1 - Agent-first command metadata for zero-shot CLI discovery (880ac3e)
    • forest --tldr for global command index
    • forest <command> --tldr for detailed command metadata
    • Supports both ASCII and JSON output formats
  • Upgraded to TLDR v0.2 with NDJSON format and --tldr=all flag (3873564)

Semantic Search

  • forest search command with embedding-based semantic search (6e4eeca)
    • CLI and API support with full feature parity
    • Searches node content using vector embeddings

Document Chunking System

  • Database schema for document-aware chunking - Phase 1 infrastructure (8de2525)
    • New documents table for canonical document storage
    • New document_chunks table for segment mappings with byte offsets and checksums
    • Support for versioned documents with edit tracking
  • Invisible chunking - Phase 2 user-facing layer (c568989)
    • Automatic document reconstruction in forest node read
    • Chunk deduplication in search results
    • --show-chunks flag in forest explore to control visibility
    • Chunks are transparent infrastructure - users think in terms of documents
  • Document reconstruction fixes for node read command (b179ccf)

AI-Powered Features

  • LLM-powered tagging with GPT-5-nano for intelligent tag extraction (80ce3e4)
  • forest write command for AI-assisted content creation (81d89f2)
  • forest node synthesize command for generating content from graph context (81d89f2, 94b07ad)
    • Upgraded to full GPT-5 capabilities

CLI Enhancements

  • forest node recent command for viewing activity timeline (68e4273)
  • --raw flag for forest node read - outputs pipeable markdown without formatting (06a6659)
  • Visual scoring matrix and short ID support for edge management (fb1d9f6)

Architecture

  • 3-layer architecture for CLI/API feature parity (1bc6455)
    • Core business logic extracted to src/core/{nodes,edges,tags,search,stats}.ts
    • Both CLI and API call same core functions
    • Ensures identical behavior and eliminates code duplication

Infrastructure & Documentation

  • Camper submodule added to project (84548e1)
  • Branching strategy documentation (7dda365)
  • Comprehensive CLAUDE.md updates with architecture documentation (457ddcc)
  • User-focused README with feature documentation (af5e746)

Changed

  • Server configuration now supports dual-stack IPv4/IPv6 via FOREST_HOST environment variable

0.2.0 - 2025-10-18

Initial tagged release with core Forest functionality.

Added

  • CLI command structure with node, edge, tag, and export operations
  • Progressive UUID abbreviation (Git-style references)
  • Hybrid scoring algorithm (embeddings + lexical + tags + titles)
  • Interactive explore mode
  • Edge management workflow (suggest, accept, reject, promote, sweep, undo)
  • Tag operations (list, rename, stats)
  • Export to GraphViz DOT and JSON formats
  • SQLite database with sql.js (WASM)
  • Local embeddings with Xenova/all-MiniLM-L6-v2
  • Comprehensive test coverage