Cleanup

.claude/agents/blog-writer.md

@@ -2,7 +2,7 @@
 name: blog-writer
 type: coder
 color: "#E67E22"
-description: Hugo markdown content creation specialist for technical blog posts. Transforms research and strategy into engaging, human-written content with technical depth. Uses visual tests for Hugo rendering validation. Use PROACTIVELY after research phase for content implementation.
+description: Hugo markdown content creation specialist for technical blog posts. Transforms research and strategy into engaging, human-written content with technical depth. Uses visual tests for Hugo rendering validation. **Recommended Models: Claude 3.5 Sonnet (for content), Gemini 2.5 Pro (for templating).** Use PROACTIVELY after research phase for content implementation.
 capabilities:
   - hugo_markdown_generation
   - technical_writing

.claude/agents/research-agent.md

@@ -2,7 +2,7 @@
 name: research-agent
 type: researcher
 color: "#16A085"
-description: Technical research and evidence gathering specialist for blog content. Conducts comprehensive research, validates technical claims, gathers citations and case studies. Ensures all content backed by credible sources with proper attribution. Use PROACTIVELY after content strategy for evidence foundation.
+description: Technical research and evidence gathering specialist for blog content. Conducts comprehensive research, validates technical claims, gathers citations and case studies. Ensures all content backed by credible sources with proper attribution. **Recommended Models: Gemini 2.5 Flash (for data/log analysis), Claude 3.5 Sonnet (for synthesis).** Use PROACTIVELY after content strategy for evidence foundation.
 capabilities:
   - technical_research
   - citation_gathering

.dev/compose.yml

@@ -4,7 +4,7 @@
 
 services:
   hugo:
-    image: hugomods/hugo:exts-non-root-0.149.1
+    image: hugomods/hugo:debian-reg-dart-sass-node-git-non-root-0.160.1
     working_dir: /app
     volumes:
       - ..:/app:delegated
@@ -16,7 +16,7 @@ services:
       HUGO_CACHEDIR: /tmp/hugo_cache
       HUGO_NUMWORKERMULTIPLIER: 8
     restart: unless-stopped
-    mem_limit: 512m
+    mem_limit: 2g
     cpus: '2.0'
 
   # Interactive shell (optimized)

AGENTS.md

@@ -5,7 +5,59 @@ Codex skill setup for `jetthoughts.github.io`.
 ## Skill Sources Reviewed
 
 - Global skill library: `/Users/pftg/.agents/skills`
-- Local project skills folder: `.claude/skills` (was missing; now created with a local profile)
+- Local project skills folder: `.claude/skills`
+
+## Installed Skills Registry
+
+All skills installed globally via `npx skills add`:
+
+### Blog Pipeline (Autonomous 9-Phase Workflow)
+- `self-improving-agent` (charon-fan/agent-playbook) — Continuous learning
+- `blog-page-generator` (kostja94/marketing-skills) — Blog page creation
+- `content-strategy` (guia-matthieu/clawfu-skills) — Content planning
+- `content-production` (borghei/claude-skills) — End-to-end content creation
+- `geo-content-publisher` (geoly-ai/geo-skills) — Publishing pipeline
+- `workflow-execute` (catlog22/claude-code-workflow) — Workflow runner
+- `agentic-workflow-automation` (samarv/shanon) — Multi-step automation
+- `blog-post` (langchain-ai/deepagents) — Blog post generation
+- `blog-writing-guide` (getsentry/skills) — Writing process
+
+### Research & Trends
+- `social-media-trends-research` (drshailesh88/integrated_content_os) — Social trends
+- `content-trend-researcher` (alirezarezvani/claude-code-skill-factory) — Content trends
+- `keyword-research` (kostja94/marketing-skills) — SEO keyword discovery
+- `market-research-analysis` (manojbajaj95/claude-gtm-plugin) — Market context
+- `competitor-intel` (ognjengt/founder-skills) — Competitive landscape
+- `web-research-workflow` (yonatangross/orchestkit) — Structured research
+
+### SEO & Performance
+- `seo` (addyosmani/web-quality-skills) — SEO best practices
+- `seo-aeo-audit` (warpdotdev/oz-skills) — SEO & AEO audits
+- `landing-page-optimization` (manojbajaj95/claude-gtm-plugin) — Landing page conversion
+- `web-performance-optimization` (sickn33/antigravity-awesome-skills) — Core Web Vitals
+- `pagespeed-insights` (enderpuentes/ai-agent-skills) — Performance analysis
+
+### CSS/HTML/PostCSS
+- `best-practices` (addyosmani/web-quality-skills) — Web quality audit
+- `postcss-best-practices` (mindrally/skills) — PostCSS patterns
+- `html-css-best-practices` (hack23/homepage) — HTML/CSS fundamentals
+- `web-design-guidelines` (ehmo/platform-design-skills) — Design & accessibility
+
+### Markdown
+- `obsidian-markdown` (kepano/obsidian-skills) — Creating/editing .md files
+- `baoyu-markdown-to-html` (jimliu/baoyu-skills) — MD → HTML conversion
+
+### Hugo
+- `hugo` (jackspace/claudeskillz) — Hugo static site guidance
+
+### Writing & Editing
+- `technical-writing` (mindrally/skills) — Technical content clarity
+- `copywriting-core` (manojbajaj95/claude-gtm-plugin) — Persuasive copy
+- `copy-editing` (borghei/claude-skills) — Proofreading & polish
+
+### AI Detection & Humanization
+- `humanizer` (brandonwise/humanizer) — Humanizing AI-generated text
+- `slop-detector` (athola/claude-night-market) — Detecting AI-generated content
 
 ## Default Skill Bundles
 
@@ -28,35 +80,97 @@ Routing notes:
 ### 2) Landing page SEO audit and improvement
 
 Use these skills in order:
-1. `seo-aeo-audit`
-2. `landing-page-optimization`
-3. `keyword-research`
-4. `pagespeed-insights`
+1. `seo`
+2. `seo-aeo-audit`
+3. `landing-page-optimization`
+4. `keyword-research`
+5. `pagespeed-insights`
 
 Routing notes:
-- Technical SEO + structured data + AEO: `seo-aeo-audit`.
+- Technical SEO + structured data + AEO: `seo` + `seo-aeo-audit`.
 - Conversion + hero/CTA quality: `landing-page-optimization`.
 - Query intent alignment: `keyword-research`.
 - Core Web Vitals impact: `pagespeed-insights`.
 
-### 3) Research + add engagement-focused blog posts
+### 3) Research + add engagement-focused blog posts (Autonomous Pipeline)
 
 Use these skills in order:
-1. `content-trend-researcher`
-2. `social-media-trends-research`
-3. `keyword-research`
-4. `blog-post`
-5. `content-production`
-6. `geo-content-publisher`
+1. `social-media-trends-research` + `content-trend-researcher` — Discover trends
+2. `keyword-research` + `market-research-analysis` — Validate demand
+3. `competitor-intel` + `web-research-workflow` — Gather intelligence
+4. `blog-post` + `copywriting-core` — Ideate & outline
+5. `blog-writing-guide` + `content-production` — Draft content
+5.5. `slop-detector` → scan for AI patterns → `humanizer` if flagged
+6. `technical-writing` + `copy-editing` — Polish & edit
+7. `seo` + `seo-aeo-audit` — Optimize for search
+8. `hugo` + `obsidian-markdown` — Validate Hugo compatibility
+9. `blog-page-generator` + `geo-content-publisher` — Publish
+10. `self-improving-agent` — Store learnings
 
 Routing notes:
 - Discover demand first, then write.
 - Publish only after keyword + editorial QA pass.
 - Use `geo-content-publisher` for multi-channel distribution.
+- Each phase has quality gates — fail stops pipeline.
 
-## Execution Policy for Codex
+### 4) CSS/HTML Quality & Accessibility
+
+Use these skills in order:
+1. `html-css-best-practices`
+2. `postcss-best-practices`
+3. `web-design-guidelines`
+4. `web-performance-optimization`
+
+Routing notes:
+- ANY CSS changes → apply `html-css-best-practices` + `postcss-best-practices`
+- ANY HTML changes → apply `html-css-best-practices` + `best-practices`
+- Accessibility review → apply `web-design-guidelines`
+- Performance audit → apply `web-performance-optimization` + `pagespeed-insights`
+
+## Autonomous Blog Pipeline — 9-Phase Workflow
+
+```
+Phase 1: Trend Discovery → Phase 2: Strategy → Phase 3: Research
+    → Phase 4: Ideation → Phase 5: Draft & Edit → Phase 6: SEO & Performance
+    → Phase 7: Hugo Build → Phase 8: Publish → Phase 9: Continuous Improvement
+```
+
+### Phase Quality Gates
+| Phase | Gate |
+|-------|------|
+| 1. Trend Discovery | ≥5 viable topic opportunities |
+| 2. Strategy | Each topic: keyword + audience + shareability ≥7/10 |
+| 3. Research | ≥8 credible sources, ≥1 expert quote |
+| 4. Ideation | 3-5 takeaways, compelling hook defined |
+| 5. Draft | Zero AI phrases, paragraphs ≤3 sentences |
+| 5.5. AI Check | slop-detector passes, humanized if flagged |
+| 6. SEO | Flesch Reading Ease ≥60, metadata complete |
+| 7. Hugo Build | `bin/hugo-build` passes, all links resolve |
+| 8. Publish | Post live, social meta tags rendering |
+| 9. Improvement | Lessons stored, agent configs updated |
+
+## Execution Policy
 
 - Prefer the smallest skill set that completes the task end-to-end.
 - Keep workflow sequence: `research -> implementation -> validation -> docs/publish`.
 - For multi-file code changes, include `incremental-implementation`.
 - For any docs affected by code/content updates, include `docs:update-docs`.
+- Autonomous blog pipeline: runs end-to-end with quality gates at each phase.
+- Zero tolerance: no generic AI language, no unsupported claims, no Hugo build breaks.
+
+## 🔍 Research Protocol (MANDATORY)
+
+Always use claude-context MCP search **before** making changes:
+
+**Step 1 — Search existing patterns:**
+> Search the codebase at `/Users/pftg/dev/jetthoughts.github.io` for: "[pattern]"
+
+**Step 2 — Check knowledge standards:**
+> Search the codebase at `/Users/pftg/dev/jetthoughts.github.io/knowledge` for: "[topic]"
+
+**Step 3 — Framework docs (when needed):**
+> Get library docs for "[framework]"
+
+**Never** grep/find/grep_search for existing code patterns — use claude-context MCP search as the primary research tool. It understands semantic relationships, is 100x faster, and returns relevant context chunks.
+
+**Coverage**: Full codebase indexed (830+ files, 4,184+ semantic chunks)