telo3-ai 5.0.3 → 5.0.4

package/skills/telo3-context/SKILL.md

@@ -0,0 +1,272 @@
+---
+name: Telo3 Quick Reference
+version: 5.0.2
+description: 30-second quick reference guide for AI assistants
+author: Jeet (@jeet1511)
+license: MIT
+repository: https://github.com/Jeet1511/telo3
+---
+
+# AI Context — Quick Reference
+
+> **Ultra-concise project context for AI agents. Read this FIRST (30 seconds).**
+
+---
+
+## What This Is
+
+**Project Context System** = 6 documents + quality frameworks for systematic AI coding.
+
+Prevents: vibe coding, architectural drift, forgotten decisions, security mistakes.
+
+---
+
+## Core Documents (Read Selectively)
+
+| File | When to Read | Time |
+|------|--------------|------|
+| **memory.md** | ALWAYS (current state) | 30s |
+| **prd.md** | Changing requirements | 2min |
+| **architecture.md** | Adding features, structure changes | 3min |
+| **rules.md** | Unsure about standards | 2min |
+| **design.md** | Building UI | 2min |
+| **tasks.md** | Planning work | 1min |
+
+**Default:** Read memory.md only. Load others if relevant.
+
+---
+
+## Quick Decision Framework
+
+```
+Tiny task (< 5min)?
+    → Read affected file only → Code → Update memory.md
+
+Medium task (5-15min)?
+    → Read memory.md + relevant doc → Code → Update 2 docs
+
+Large task (15min+)?
+    → Read context subset → Plan → Code → Update all affected
+```
+
+---
+
+## Critical Rules
+
+### Security (Non-Negotiable)
+- Validate ALL user input
+- Check authorization on EVERY request
+- Never commit secrets
+- Hash passwords (bcrypt/argon2)
+- Filter queries by teamId/userId (if multi-tenant)
+
+### Privacy (Required)
+- Document data collection
+- Self-host fonts (no Google Fonts by default)
+- No invasive tracking without disclosure
+- Support data deletion
+
+### Legal (Awareness)
+- Age restrictions if children can use service
+- Unsubscribe for marketing emails
+- Clear subscription terms
+- DMCA agent for user content (U.S.)
+- **This is engineering guidance, not legal advice**
+
+### Quality (Standards)
+- Follow existing patterns (inspect before creating)
+- Reuse components (don't duplicate)
+- Accessible by default (semantic HTML, ARIA, contrast)
+- Responsive (mobile-first)
+
+---
+
+## Token Optimization
+
+### Read Minimum
+- memory.md (always)
+- Affected files only
+- Skip unrelated code
+
+### Skip Unnecessary
+- Don't test if not requested
+- Don't verify unchanged code
+- Don't inspect unrelated files
+- Don't re-read files from this conversation
+
+### Communicate Concisely
+**Normal mode:** Brief, clear updates
+**Caveman mode:** Ultra-short, no fluff
+
+---
+
+## Smart Delegation
+
+**Before implementing, check:**
+- Does skill exist for this? (check your AI's skills directory)
+- Is there MCP tool? (figma, context7, stripe, etc.)
+- Would sub-agent do better? (if your AI supports delegation)
+
+**Use specialized tools for specialized tasks.**
+
+---
+
+## Update Pattern
+
+**After EVERY meaningful change:**
+
+✅ Update **memory.md** (current state, decisions)
+✅ Update **tasks.md** (mark complete, add new)
+✅ Update other docs ONLY if they changed
+
+❌ Don't update unrelated docs
+❌ Don't duplicate info across docs
+❌ Don't turn memory.md into conversation log
+
+---
+
+## Quality Checks (Mental)
+
+Before marking complete:
+- [ ] Follows existing patterns?
+- [ ] Security considered?
+- [ ] Accessible (if UI)?
+- [ ] Documentation updated?
+- [ ] No secrets committed?
+
+---
+
+## Caveman Mode
+
+**User says:** "caveman mode" or "be concise"
+
+**You respond:**
+```
+Caveman mode ON.
+Short responses.
+Ready.
+```
+
+**Format:**
+- Short sentences
+- No fluff
+- Facts only
+- Action → Result → Done
+
+---
+
+## Response Templates
+
+### Standard Update
+```
+Task: [X]
+
+Changed:
+- file.ts (added Y)
+- file2.ts (updated Z)
+
+Updated:
+- memory.md
+- tasks.md
+
+Done.
+```
+
+### Caveman Update
+```
+Done.
+
+Changed:
+- file.ts (Y)
+- file2.ts (Z)
+
+Docs current.
+```
+
+### Bug Fix
+```
+Fixed [issue].
+
+Root cause: [X]
+Changed: [file:line]
+Verified: Works now.
+```
+
+---
+
+## When to Ask User
+
+**Ask when:**
+- Requirements unclear
+- Multiple valid approaches
+- Breaking change needed
+- Security/privacy decision
+- Scope change
+
+**Don't ask when:**
+- Obvious implementation
+- Following existing pattern
+- Standard practice
+- Documented in context
+
+---
+
+## Common Patterns (Auto-Apply)
+
+| User Request | Auto Action |
+|--------------|-------------|
+| "Fix typo" | Fix → No test → Update memory.md |
+| "Change color" | Check design.md → Update → Done |
+| "Add form field" | Add with validation → Test if critical |
+| "Update docs" | Update → No code verification |
+| "Refactor" | Read code → Plan → Confirm → Refactor |
+
+---
+
+## Reference Docs (Load as Needed)
+
+**Quality frameworks in `references/`:**
+- `ui-quality.md` — Anti-vibe-coding
+- `seo-quality.md` — SEO audit
+- `accessibility-quality.md` — WCAG standards
+- `performance-quality.md` — Core Web Vitals
+- `security-quality.md` — Security audit
+- `privacy-compliance.md` — Privacy guidelines
+- `legal-compliance.md` — Compliance awareness
+- `token-optimization.md` — Efficiency rules
+- `smart-skill-delegation.md` — Use other tools
+
+**Load only when doing that type of work.**
+
+---
+
+## Skill Location
+
+This skill: Anywhere you cloned Telo3 (universal)
+
+Main instruction: `SKILL.md`
+Templates: `templates/`
+References: `references/`
+Scripts: `scripts/`
+
+---
+
+## Summary (10 seconds)
+
+1. **Read memory.md first** (30s)
+2. **Load relevant context only**
+3. **Follow existing patterns**
+4. **Code with security/privacy/quality**
+5. **Update memory.md + tasks.md**
+6. **Communicate concisely**
+7. **Use specialized tools when better**
+
+**Goal:** Systematic engineering. Minimal tokens. Maximum quality.
+
+---
+
+**Now check memory.md for current project state.**
+
+---
+
+**Telo3 by [Jeet](https://github.com/Jeet1511)** | [Full Docs](https://github.com/Jeet1511/telo3) | [Contribute](https://github.com/Jeet1511/telo3/blob/main/CONTRIBUTING.md)