Skip to content

changelog.md

Latest commit

 

History

History
100 lines (80 loc) · 7.34 KB

changelog.md

File metadata and controls

100 lines (80 loc) · 7.34 KB

Changelog

All notable changes to Clawsight.

[0.7.0] — 2026-03-31

Added — career-sim and 4-Skill Chain

  • career-sim v1.0.0: Divergent career path simulator. Generates 3-5 genuinely different career trajectories based on profile data, upstream analysis, and market signals. Comparison matrix across 8 dimensions (income trajectory, AI-proof score, skill leverage, time to impact, reversibility, market demand, compound advantage fit, risk level). Trade-off analysis with points of no return. Decision framework by priority (stability, growth, AI-readiness) with hedge strategies.
  • career-sim README.md: User-facing documentation for the career path simulator.
  • Cross-skill data passing: New CAREER_SIM_OUTPUT block with chosen path, thesis, year-1 focus, key skills needed, risk factors, and AI alignment level.

Changed

  • tech-compass updated to accept career-sim output as upstream input. When career-sim chosen_path is available, all recommendations (skill quadrant, learning routes, action plan) align to the selected trajectory. New error handling for missing career-sim data.
  • scene-skills-protocol.md updated for 4-skill chain: career-mirror → tech-spectrum → career-sim → tech-compass. Added career-sim to data flow diagram, slash command table, and chain invocation section.
  • README.md updated to reflect 4-skill chain (career-mirror → spectrum → sim → compass). Version badge updated to 0.7.0. Architecture diagram includes career-sim. Roadmap updated with v0.7 checked and v0.8/v0.9/v1.0 renumbered.

[0.6.0] — 2026-03-31

Added — Career Intelligence Chain

  • career-mirror v2.0: Rewritten to focus on pure introspection. New Advantage Verification Matrix with triple-source cross-validation (Declared × Behavioral × Third-party). 4 report sections: Career Arc, Compound Advantage Analysis, Behavioral Truth, Blind Spot Map. Strict scope boundary — observation only, no direction prescriptions.
  • tech-spectrum v1.0: AI disruption positioning skill. Five-level AI Spectrum (AI-vulnerable → AI-adjacent → AI-augmented → AI-native → AI-shaping). Three-layer analysis: AI Exposure → AI Readiness → Trend Intersection. References docs/ai-trends.md for 8-track trend data.
  • tech-compass v1.0: Action planning endpoint of the career chain. Skill Quadrant Matrix (Your-Level × Market-Demand), AI Skill Layer Assessment (L0-L4), personalized Learning Routes, 30-60-90 Day Action Plan with success criteria, Risk & Adaptation with disruption timeline.
  • Cross-skill data passing protocol: HTML comment blocks with YAML (<!-- CAREER_MIRROR_OUTPUT ... -->, <!-- TECH_SPECTRUM_OUTPUT ... -->) for structured data flow between skills.
  • Three operating modes: Enhanced (profile + upstream outputs), Rich (profile only), Lite (no profile).
  • docs/scene-skills-protocol.md — Cross-skill interaction rules, data flow, invocation patterns
  • docs/skill-layers.md — AI Skill Layers L0-L4 framework with criteria, evidence signals, and career phase recommendations
  • docs/ai-trends.md — Comprehensive AI development timeline: 130+ milestones across 8 tracks (Agent & Toolchain, AI-Native Dev, Vertical AI, Multimodal, Safety & Governance, Infrastructure, Data Engineering, Hardware/Edge)
  • README.md for each new Scene Skill (skills/tech-spectrum/README.md, skills/tech-compass/README.md)

Changed

  • career-mirror narrowed from career direction analysis to pure introspection (direction suggestions moved to tech-compass)
  • architecture.md updated with Career Intelligence Chain diagram, cross-skill data flow, MCP Enhancement Path, and trust hierarchy fix (added Third-party 0.8 weight)
  • README.md updated with Scene Skills section, three operating modes, cross-skill data flow, and new documentation links
  • examples/prompts/tech-compass.md updated to redirect to full Scene Skill implementation

Architecture

  • Defined MCP Enhancement Path: Phase 1 (Pure Skill) → Phase 2 (MCP tools for real-time data) → Phase 3 (structured data layer)
  • Scene Skill size budget: < 6KB per SKILL.md, with methodology docs extracted to docs/

[0.5.0] — 2026-03-30

Added — Layer 3: Potential Discovery

  • /clawsight potential command: Industry trend search × compound advantage mapping × opportunity gap analysis
  • Dialogue-Based Profile Enrichment (Step 8): Passive detection of new user info during normal conversation, non-intrusive update suggestions
  • Profile Evolution Tracking: Records skill shifts, activity changes, and growth patterns across refreshes
  • career-mirror Scene Skill: Independent SKILL.md for career direction analysis, published to ClawHub (skills/career-mirror/)
  • Rich Mode (with Clawsight profile) and Lite Mode (without)

Changed

  • Step 7b expanded with structured 4-part potential analysis process
  • Modularized SKILL.md: detailed templates, schemas, and scoring moved to docs/ for size optimization
  • SKILL.md reduced from 25,827B → 18,205B (29% smaller) while adding v0.4+v0.5 features

[0.4.0] — 2026-03-30

Added — Layer 2: Insight Deepening

  • LinkedIn Recommendations Parser (source_linkedin_zip): Parse recommendations.json from LinkedIn export for third-party endorsements
  • /clawsight refresh command: Re-fetch all previously imported sources, diff against stored data, staleness check (>90 days)
  • 5 Structured Insight Types: Hidden Strengths, Behavioral-Declarative Gaps, Blind Spots, Compound Advantages, Evolution Signals
  • Behavioral Pattern Analysis in source_github: Coding schedule (morning/night/consistent), consistency score, collaboration ratio
  • docs/schema.md — Full canonical extraction schema (extracted from SKILL.md)
  • docs/scoring.md — Detailed scoring methodology
  • docs/templates.md — Output templates for USER.md, MEMORY.md, preview, reports

[0.3.0] — 2026-03-30

Added — Multi-Source Intelligence Engine

  • SKILL.md v0.3: Complete rewrite as Pure Skill (no TypeScript runtime needed)
  • 7-Step Pipeline: Identify → Fetch → Parse → Cross-Source Reconciliation → Validate → Preview → Write
  • Step 3.5 Cross-Source Reconciliation: 5-type conflict detection, source-domain trust matrix, contradictions-as-insights
  • Three-Layer Intelligence: Profile (画像构建) → Insight (当下洞察) → Potential (潜力发掘)
  • GitHub Deep Parser: 4 endpoints (profile, repos, README, events)
  • Brand: Mantis Shrimp 🦐 — "See what you can't see about yourself"
  • docs/architecture.md, docs/user-journey.md
  • examples/sample-output/ — USER.md and MEMORY.md samples
  • examples/prompts/ — career-mirror and tech-compass prompt templates

Changed

  • TypeScript CLI archived to legacy/ (still functional for reference)
  • Repo renamed from claw-life-import to clawsight
  • README.md rewritten for Pure Skill focus

[0.2.0] — 2026-03-28

Added

  • Full SKILL.md with 6-step pipeline
  • Two-layer architecture: Import Engine + Memory Sync Engine
  • Evidence & provenance tracking system
  • 7 source parsers (website, GitHub, JSON Resume, LinkedIn export/URL, PDF, plain text)
  • Privacy filter with L0-L3 classification
  • Dual scoring: Profile Coverage + Assistant Understanding
  • Preview-first confirmation flow

[0.1.0] — 2026-03-25

Added

  • Initial TypeScript CLI (claw-life-import)
  • Basic resume import pipeline
  • PDF, JSON, and text parser support
  • USER.md and MEMORY.md writer