Skip to content

Latest commit

 

History

History
105 lines (83 loc) · 7.05 KB

File metadata and controls

105 lines (83 loc) · 7.05 KB

Changelog

1.3.0

  • Ada character system — Ada now has a Samantha-like personality with emotional memory, self-reflection, and growth across sessions
  • Obsidian CLI integration — New _util-obsidian skill wraps Obsidian CLI (v1.12+) for vault search, backlinks, tags, and file operations
  • Context-enriched assistants — Calendar, Projects, and Relationships assistants now pull related vault content via Obsidian CLI for richer context
  • Added Ada's semantic memory (Semantic/Ada/memory.md) for emotional state and relationship history
  • Added learning references for Ada's post-ritual self-reflection at all 4 timescales
  • Added plan/reflect references for Ada's ritual integration
  • Updated scaffold with Ada's memory and output directories
  • Added worktree cleanup order rule to CLAUDE.md

1.2.0

  • Parallel assistant execution — Ada now runs assistants as parallel background agents for 70-80% faster rituals
  • Draft-and-fill architecture — Assistants output drafts with HTML comment placeholders, Ada orchestrates questions and fills answers
  • Intelligent question ordering — Ada orders questions by priority (context → strategy → tactics → capture) for natural conversation flow
  • Graceful degradation — Failed agents don't block ritual completion; partial results still usable
  • Feature flag support — Control parallel execution via vault/00_Brain/Systemic/Config/ada.yaml
  • Updated all 7 assistants (goals, calendar, journal, achievements, relationships, projects) for draft mode
  • Updated all plan and reflect rituals across all 4 timescales
  • Added comprehensive documentation for parallel execution and draft mode
  • Added error handling: timeouts, failures, malformed placeholders

0.9.0

  • Absorbed init skill into Ada — /ada setup replaces /init, Ada is now the central UX for all human interactions
  • Merged brief assistant into compose phase — brief synthesis happens during composition with access to all assistant outputs
  • Split templates into Ada base + assistant sections — 19 assistant-owned templates, 4 Ada base templates with compose order
  • Added variance-based reflect flow — each assistant compares plan output to actual capture, returns findings to Ada for coaching
  • Moved section order to Ada's systemic memory — vault/00_Brain/Systemic/Ada/section-order.md stores Ada's preferences
  • Renamed ada.yamlassistants.yaml for user-configurable assistant settings
  • Moved protocol specs to _specs/ and compose to ada/references/compose/ (never were assistants)
  • Removed _assistant-brief and init skills entirely
  • Updated all documentation and cross-references

0.8.0

  • Restructured documentation for new user journey — 5 top-level sections: Overview → Configuration → Ada → Assistants → Reference
  • Added Overview section with "What is 2bd?", "What You Get", and "Quick Start" pages
  • Created Configuration section consolidating installation, settings, and hooks documentation
  • Promoted Assistants to top-level section with overview and 7 assistant pages
  • Consolidated development docs into Reference — renamed "skills" to "creating-custom-assistants", "rituals" to "creating-custom-rituals"
  • Merged tracks documentation into creating-custom-assistants (assistant matrix, coach system)
  • Moved skill writing style and GitHub workflow labels to CLAUDE.md
  • Fixed all internal documentation links
  • Removed obsolete docs/setup/ and docs/development/ directories

0.7.0

  • Implemented Ada-centric skill architecture — Ada owns all logic, entity skills are thin dispatchers
  • Moved entity actions into ada/references/{entity}/{action}.md (project/create, project/archive, person/onboard)
  • Introduced prefix scheme: (none)=user-invokable, _assistant-*=internal assistants, _util-*=utilities
  • Renamed utilities: _resolve-date_util-resolve-date, _append-changelog_util-changelog
  • Simplified aliases: planningplan, reflectionreflect
  • Updated ritual command examples in ada/how-it-works.md to match renamed commands
  • Updated documentation with new skill architecture and prefix conventions

0.6.0

  • Restructured docs with Ada-first navigation — Ada is now the top-level entry point with domain assistants beneath
  • Created dedicated documentation pages for all 7 domain assistants (brief, goals, calendar, journal, achievements, relationships, projects)
  • Absorbed GPS methodology into ada/goals.md and rituals reference into ada/how-it-works.md
  • Removed overview/concepts.md, overview/gps-methodology.md, overview/how-it-works.md, reference/rituals.md
  • Updated all cross-references across documentation

0.5.0

  • Refactored init to be Ada-orchestrated — init is now a thin wrapper that invokes Ada for the setup experience
  • Added awakening script to Ada with Samantha-style personality (curious, self-discovering, warm but not performative)
  • Added setup and profile actions to Ada for first-run and profile updates
  • Renamed user-profile.mdhuman.md and ai-personality.mdada.md in scaffold templates
  • Removed Monthly from scaffold (timescale no longer exists)
  • Updated docs to reflect new file names and /ada profile command

0.4.0

  • Added timescale declarations to assistant frontmatter — each assistant's SKILL.md now declares timescales.plan and timescales.reflect arrays
  • Ada reads frontmatter directly to determine which assistants to invoke at each timescale
  • Added _assistant-protocol/references/timescales.md documenting the frontmatter contract
  • Replaced hardcoded timescale table in ritual-flow.md with reference to frontmatter declarations

0.3.0

  • Renamed review rituals to reflection rituals (ritual-review-*ritual-reflection-*) for semantic alignment
  • Renamed Reflection phase to Learning phase across all rituals to avoid terminology collision
  • Updated all documentation, templates, and scaffold files to use new terminology

0.2.0

  • Added timescale coaches with configurable personality — 12 coaching files organized as Coaching/{domain}/{timescale}.md where domain is leadership/planning/review and timescale is daily/weekly/quarterly/yearly
  • Added coach personality configuration in ai-personality.md with Character, Ego, Challenge Level, and Celebration Style settings
  • Added Coaching Preferences section to init wizard with 6 coach archetypes: work wife, brutally honest sparring partner, executive assistant with career ambitions, wise mentor, emo teenager, drill sergeant
  • Updated all 8 rituals to load timescale-specific coaching files instead of flat domain files
  • Removed old flat coaching files (leadership.md, planning.md, review.md) in favor of new directory structure
  • Fixed CLAUDE.md timescale architecture section to include coaching file locations
  • Fixed CLAUDE.md agent notes to document vault symlink

0.1.0

  • Initial release with daily, weekly, quarterly, and yearly planning and review rituals
  • Vault scaffold with Systemic templates, Coaching guides, and Directive files
  • Init wizard for fresh install, reconnect, and profile update
  • Project and people management skills