Cross-source career intelligence for the AI era — know yourself, find your position, take action.
Your resume says "Java lead" but your GitHub is 90% Go. Three LinkedIn recommenders highlight your mentoring, yet your resume never mentions leadership. You sit at the intersection of payment systems × distributed architecture × global ops — a combination held by fewer than 2% of senior engineers. And in an AI revolution reshaping every career, you have no idea whether your next move should be doubling down or pivoting.
These insights are invisible from any single source. They only emerge when you look across data streams — and map them against the forces reshaping your industry.
Clawsight does exactly that. Named after the Mantis Shrimp 🦐 — the creature with 16 types of color receptors that sees dimensions invisible to all other species.
# 1. Install
/install clawsight
# 2. Import your data (the more sources, the sharper the insight)
/clawsight resume.pdf
/clawsight https://github.com/yourusername
/clawsight linkedin.zip
# 3. Run the Career Intelligence Chain
/career-mirror # Who am I really? (introspection)
/tech-spectrum # Where do I stand in the AI revolution? (positioning)
/career-sim # What paths are open to me? (simulation)
/tech-compass # What do I do next? (action plan)No data? Every skill works in Lite Mode too — just with interactive Q&A instead of cross-source depth.
This is Clawsight's core value proposition: a four-skill pipeline that takes you from self-knowledge to concrete action.
🪞 career-mirror 🌈 tech-spectrum 🔮 career-sim 🧭 tech-compass
───────────────── → ───────────────── → ───────────────── → ─────────────────
"Who am I really?" "Where do I stand "What paths are "What do I do
in the AI era?" open to me?" next?"
▸ Career Arc ▸ Five-Level Spectrum ▸ 3-5 Divergent Paths ▸ Skill Quadrant Matrix
▸ Advantage Verify ▸ AI Exposure Analysis ▸ Comparison Matrix ▸ AI Skill Layer (L0→L4)
▸ Behavioral Truth ▸ Trend × Profile ▸ Trade-off Analysis ▸ Learning Routes
▸ Blind Spot Map ▸ Opportunity Windows ▸ Decision Framework ▸ 30-60-90 Day Plan
Each skill automatically passes structured data downstream, so the chain builds on itself:
| Skill | Command | Input | Output |
|---|---|---|---|
| career-mirror | /career-mirror |
Clawsight profile | Verified advantages, behavioral patterns, blind spots |
| tech-spectrum | /tech-spectrum |
Profile + career-mirror | AI spectrum position, exposure analysis, opportunity windows |
| career-sim | /career-sim |
Profile + both upstreams | 3-5 divergent career paths, comparison matrix, trade-offs |
| tech-compass | /tech-compass |
Profile + all upstreams | Skill quadrant, AI skill layer, action plan for chosen path |
Every Scene Skill adapts to what's available:
| Mode | When | Experience |
|---|---|---|
| Enhanced | Profile + upstream skill outputs | Full chain with cross-validated intelligence |
| Rich | Clawsight profile only | Profile-driven analysis, no upstream data |
| Lite | Nothing imported | Interactive Q&A — still useful, just less precise |
The engine behind everything. When you import 2+ sources, Clawsight doesn't just merge — it cross-references:
Resume says: "Java Lead, 8 years"
GitHub shows: 47 repos, 92% Go, 3 CNCF contributions
LinkedIn recs: "exceptional mentoring", "natural leader"
Clawsight sees:
→ Behavioral-Declarative Gap: identity shifted to Go, resume tells Java story
→ Hidden Strength: leadership evidenced by others but never self-claimed
→ Compound Advantage: payments × distributed × global = rare triple stack
Five conflict types detected (factual, temporal, emphasis, omission, granularity). Four auto-resolved. Factual contradictions escalated to user. Contradictions become insights, not errors.
Not all data is equal:
| Source Type | Trust Weight | Example |
|---|---|---|
| Behavioral | 0.9 | GitHub commits, contribution patterns |
| Third-party | 0.8 | LinkedIn recommendations, endorsements |
| Declarative | 0.7 | Resume claims, self-descriptions |
| Inferred | 0.5 | Cross-reference deductions |
| Source | Command | What It Captures |
|---|---|---|
| Resume (PDF/text) | /clawsight resume.pdf |
Career narrative, declared skills, achievements |
| GitHub | /clawsight https://github.com/user |
Real tech stack, contribution patterns, coding rhythm |
| LinkedIn export | /clawsight linkedin.zip |
Recommendations, endorsements, network signals |
| Personal website | /clawsight https://yoursite.com |
Self-presentation, projects, writing style |
| JSON Resume | /clawsight resume.json |
Structured profile data |
LinkedIn: Export via Settings → Get a copy of your data. See LinkedIn Guide for steps.
| Command | Description |
|---|---|
/clawsight insight |
Hidden strengths, blind spots, behavioral-declarative gaps |
/clawsight potential |
Compound advantages × industry trends mapping |
/clawsight score |
Profile completeness and understanding level |
/clawsight refresh |
Re-fetch all sources, track profile evolution over time |
Sources Memory Scene Skills
───────┈ ────── ────────────
Resume ─┐ USER.md 🪞 career-mirror
GitHub ─┤ Parse → MEMORY.md → 🌈 tech-spectrum
LinkedIn┤ Reconcile → memory/projects/* 🔮 career-sim
Website ┘ Write → 🧭 tech-compass
📝 writing-voice *
⛔ Privacy Preview 📚 learning-path *
before any write 👤 stakeholder-briefer *
* Planned
Pure Skill architecture: No runtime. No dependencies. No compiled code. Just SKILL.md files interpreted by your AI agent. Works with any OpenClaw-compatible environment (Claude Code, etc.).
- Read-only Scene Skills: Scene Skills consume profile data but never write to it. Data integrity preserved.
- Cross-skill data passing: HTML comment blocks with structured YAML, appended to reports. Downstream skills parse them silently.
- Graceful degradation: Every skill works without a profile. More data = sharper insight, but zero data ≠ zero value.
- < 6KB per Scene Skill: Each SKILL.md is self-contained and stays under the size budget. Methodology and reference data live in
docs/.
See docs/architecture.md for the full technical deep dive.
Clawsight ships with a structured AI development timeline — the data backbone for tech-spectrum's positioning analysis:
- 130+ milestones across 8 tracks: Agent & Toolchain, AI-Native Dev, Vertical AI, Multimodal, Safety & Governance, Infrastructure, Data Engineering, Hardware/Edge
- Each track with: development phases, key milestones, acceleration rating, career impact assessment
- Cross-track fusion analysis with scarcity ratings for intersection opportunities
- Updated quarterly; designed for community contribution
See docs/ai-trends.md for the full timeline.
Clawsight is designed to evolve from pure prompt intelligence to tool-augmented intelligence:
| Phase | What | Status |
|---|---|---|
| Phase 1: Pure Skill | LLM general knowledge + user profile | ✅ Current |
| Phase 2: MCP Tools | Web search for real-time trends, job market APIs for demand validation | 🔜 Next |
| Phase 3: Data Layer | Structured databases, skill taxonomies, market indices | 📋 Planned |
Every data claim in output is tagged [data-based], [general-knowledge], or [real-time] so users always know the source.
| Doc | Content |
|---|---|
| Architecture | System design, pipeline detail, data flow |
| Scene Skills Protocol | How Scene Skills interact, data passing format, mode detection |
| AI Trends | 8-track AI development timeline (130+ milestones) |
| Skill Layers | AI Skill Layers L0→L4 framework with assessment criteria |
| Schema | Canonical data extraction schema |
| Scoring | Profile completeness scoring methodology |
| Templates | Output templates for reports and memory files |
| User Journey | Interaction lifecycle and onboarding flow |
| LinkedIn Guide | Step-by-step LinkedIn data export |
| Changelog | Version history |
- v0.3 — Multi-source engine + Pure Skill rewrite
- v0.4 — Insight deepening + LinkedIn recommendations + refresh
- v0.5 — Potential discovery + dialogue enrichment + career-mirror v1
- v0.6 — Career Intelligence Chain (career-mirror v2 + tech-spectrum + tech-compass)
- v0.7 — career-sim + 4-skill chain (mirror → spectrum → sim → compass)
- v0.8 — MCP Phase 2: real-time trend data + job market validation
- v0.9 — writing-voice + learning-path Scene Skills
- v1.0 — OpenClaw profile standard + community skill marketplace
Clawsight is an open-source project. Here's how you can contribute:
Build a Scene Skill — The highest-impact contribution. Create a new skill under skills/ that consumes Clawsight profile data. Reference: career-mirror, protocol: scene-skills-protocol.md.
Maintain AI Trends — Help keep docs/ai-trends.md current with new milestones, updated phase assessments, and career impact analysis.
Add a Source Parser — Propose new sources (Stack Overflow, Dribbble, Behance, etc.) by opening an issue.
Sharpen Reconciliation — The cross-source heuristics in SKILL.md improve with real-world edge cases. If you find a conflict type it handles poorly, report it.
Fix Bugs & Docs — Typos, unclear instructions, missing edge cases — all PRs welcome.
git checkout -b feature/your-contribution
# Make changes, then open a PRMIT — use it, modify it, ship it.