@frase/agent-skills 0.3.0

package/frase-playbook-automation/SKILL.md

@@ -0,0 +1,181 @@
+---
+name: frase-playbook-automation
+description: Automate multi-step content workflows using Frase playbooks, templates, and content rules — chain research to publish with quality gates. Works via MCP or CLI.
+---
+
+# Frase Playbook Automation
+
+This skill teaches you how to automate content workflows using Frase's playbooks, templates, and content rules — chain multi-step operations with quality gates and human review points.
+
+**Works with:** Frase MCP server (`@frase/mcp-server`) OR Frase CLI (`@frase/cli`)
+
+## Prerequisites
+
+- Frase MCP server connected OR `frase` CLI installed (`npx @frase/cli`)
+- `FRASE_API_KEY` environment variable set (get from https://next.frase.io/settings/api)
+- A Frase site with content (check: `list_sites` / `frase sites list`)
+- Scale plan or higher (for playbook features)
+
+## Concepts
+
+### Playbooks
+Multi-step workflow chains: research → brief → generate → optimize → rule check → human review → publish. Steps run sequentially, passing outputs between them via `{{outputs.briefId}}`.
+
+### Templates
+Structured blueprints applied to briefs: section definitions, boilerplate blocks, field defaults, word count targets. Ensure consistent content structure across articles.
+
+### Content Rules
+Quality gates with 9 condition types: score thresholds (SEO/GEO), term compliance (brand terminology), word count, meta tags, link counts, section requirements, regex patterns, competitor mention filters, and verify chip limits.
+
+### Rule Sets
+Groups of rules applied together. Max 3 enabled rule sets per organization. Evaluate content against all rules in a set before publishing.
+
+## Workflow
+
+### Step 1: Set Up Content Rules
+
+Define quality standards your content must meet before publishing.
+
+**Via MCP:**
+```
+1. Use `list_rule_sets` to see existing rule sets
+2. Use `get_rule_set` to inspect a specific rule set's rules and configuration
+3. Use `evaluate_rules` to check content against a rule set
+4. Note: Rule set creation/editing is done via the Frase UI or CLI, not MCP
+5. Rule types:
+   - score_threshold: minimum SEO or GEO score (e.g., { type: "seo_score", min: 70 })
+   - term_compliance: required/forbidden terms
+   - word_count: min/max word count
+   - meta_tags: require title, description, etc.
+   - link_count: min internal/external links
+   - section_requirements: required H2 sections
+   - regex_pattern: custom pattern matching
+   - competitor_mentions: filter competitor brand names
+```
+
+**Via CLI:**
+```bash
+frase rules list --site-id <id>
+frase rules set create "Blog Quality Standards" --site-id <id>
+frase rules create --rule-set-id <id> --type score_threshold --config '{"metric":"seo_score","min":70}'
+```
+
+### Step 2: Create or Choose a Template
+
+Templates ensure consistent content structure.
+
+**Via MCP:**
+```
+1. Use `list_templates` to see available templates
+2. Use `apply_template` to apply a template when creating a brief
+3. Or use `extract_template` from an existing high-performing brief to reuse its structure
+```
+
+**Via CLI:**
+```bash
+frase template list --site-id <id>
+frase template get <template-id>
+frase template apply <template-id> --brief-id <brief-id>
+frase template extract --brief-id <brief-id>   # create template from existing brief
+```
+
+### Step 3: Create a Playbook
+
+Chain multiple steps into an automated workflow.
+
+**Via MCP:**
+```
+1. Use `list_playbooks` to see existing playbooks
+2. Use `get_playbook` to review a playbook's steps
+3. Playbook steps can include:
+   - research: Run research on a topic
+   - brief: Create a brief (with optional template)
+   - generate: Generate content from brief
+   - optimize: Run optimization
+   - rule_check: Evaluate content against rule set
+   - human_review: Pause for human approval
+   - publish: Publish to CMS
+4. Steps chain via output references: {{outputs.briefId}}, {{outputs.contentId}}
+```
+
+**Via CLI:**
+```bash
+frase playbook list --site-id <id>
+frase playbook get <playbook-id>
+```
+
+### Step 4: Run the Playbook
+
+**Via MCP:**
+```
+1. Use `run_playbook` with:
+   - playbook_id: the playbook to execute
+   - topic: the target keyword or topic (optional)
+   - content_id: existing content ID to run against, e.g. for optimization playbooks (optional)
+2. Poll `get_playbook_run_status` for progress
+3. Each step completes before the next begins
+4. Human review steps pause until approved
+5. If a step fails, onFail config determines behavior:
+   - "retry_previous": retry the failed step
+   - "stop": halt the playbook
+   - "skip": skip and continue
+```
+
+**Via CLI:**
+```bash
+frase playbook run <playbook-id> --topic "keyword research tools" --site-id <id>
+frase playbook status <run-id>           # poll progress
+frase playbook status <run-id> --compact  # short status for polling
+```
+
+### Step 5: Monitor and Iterate
+
+**Via MCP:**
+```
+1. Use `get_playbook_run_status` to check each step's outcome
+2. Review any rule failures — they indicate content quality issues
+3. Use `provide_playbook_input` to respond to human review steps or provide required input
+4. Note: Cancelling runs is available via CLI (`frase playbook cancel`) but not via MCP
+```
+
+**Via CLI:**
+```bash
+frase playbook status <run-id>
+frase playbook cancel <run-id>
+```
+
+### Step 6: Evaluate Content Against Rules (Standalone)
+
+You can evaluate any content against rules without running a full playbook:
+
+**Via MCP:**
+```
+1. Use `evaluate_rules` with content_id and rule_set_id
+2. Review pass/fail for each rule
+3. Fix failing rules before publishing
+```
+
+**Via CLI:**
+```bash
+frase rules evaluate --content-id <id> --rule-set-id <rule-set-id>
+```
+
+## Example Playbook Flow
+
+```
+┌──────────┐    ┌───────┐    ┌──────────┐    ┌──────────┐    ┌────────────┐    ┌─────────┐
+│ Research  │ →  │ Brief │ →  │ Generate │ →  │ Optimize │ →  │ Rule Check │ →  │ Publish │
+│          │    │       │    │          │    │ (≥70)    │    │ (pass all) │    │         │
+└──────────┘    └───────┘    └──────────┘    └──────────┘    └────────────┘    └─────────┘
+                                                  ↑               │
+                                                  └── retry ──────┘ (if score < 70)
+```
+
+## Key Principles
+
+1. **Rules before publishing** — never publish content that fails quality rules
+2. **Templates for consistency** — reuse templates across similar content types
+3. **Extract from winners** — create templates from your best-performing content
+4. **Human review for high-stakes** — add review gates for commercial and legal content
+5. **Start simple** — begin with 3-4 step playbooks, add complexity as needed
+6. **Monitor run failures** — failed steps reveal process issues to fix

package/frase-seo-audit/SKILL.md

@@ -0,0 +1,145 @@
+---
+name: frase-seo-audit
+description: Run and interpret SEO/GEO audits using Frase — site crawls, scoring, site health, cannibalization detection, and prioritized fix workflows. Works via MCP or CLI.
+---
+
+# Frase SEO & GEO Audit
+
+This skill teaches you how to run and interpret site audits using Frase — covering SEO scoring, GEO scoring, site health, cannibalization, and internal link analysis.
+
+**Works with:** Frase MCP server (`@frase/mcp-server`) OR Frase CLI (`@frase/cli`)
+
+## Prerequisites
+
+- Frase MCP server connected OR `frase` CLI installed
+- `FRASE_API_KEY` set (get from https://next.frase.io/settings/api)
+- A site registered in Frase (check: `list_sites` / `frase sites list`)
+
+## Workflow
+
+### Step 1: Quick Site Health Check
+
+Before committing to a full audit, get the site health overview.
+
+**Via MCP:**
+```
+1. Use `get_site_health` with site_id to get:
+   - Overall health score (0-100)
+   - Issue counts by severity
+   - Page statistics
+```
+
+**Via CLI:**
+```bash
+frase health overview --site-id <id>
+```
+
+### Step 2: Check for Cannibalization
+
+Pages competing for the same keywords hurt each other's rankings.
+
+**Via MCP:**
+```
+1. Use `get_cannibalization` with site_id
+2. Review keyword groups where multiple pages compete
+```
+
+**Via CLI:**
+```bash
+frase health cannibalization --site-id <id>
+```
+
+### Step 3: Run a Full Site Audit
+
+**Via MCP:**
+```
+1. Use `start_audit` with site_id and type: "full"
+2. Poll `get_job_status` until the audit completes
+3. Use `get_audit` to retrieve the full results
+```
+
+**Via CLI:**
+```bash
+frase audit start --site-id <id> --type full
+frase job status <audit-id>             # poll until ready
+frase audit get <audit-id>
+```
+
+### Step 4: Review Audit Results
+
+```
+1. Focus on issues ranked by estimated traffic impact
+2. Check SEO and GEO scores per page
+3. Export for spreadsheet analysis:
+   - MCP: `export_audit` with format: "csv"
+   - CLI: frase audit export <id> --format csv
+```
+
+### Step 5: Analyze Internal Links
+
+**Via MCP:**
+```
+1. Use `get_link_graph` with site_id
+2. Identify orphan pages (no internal links pointing to them)
+3. Find most-linked and least-linked pages
+```
+
+**Via CLI:**
+```bash
+frase health links --site-id <id>
+```
+
+### Step 6: Prioritize Fixes
+
+**Fix content issues before technical issues** — content fixes typically have higher ROI.
+
+Priority order:
+1. **High-traffic pages with low SEO/GEO scores** — immediate impact
+2. **Cannibalization conflicts** — consolidate or differentiate competing pages
+3. **Pages with declining rankings** — prevent further traffic loss
+4. **Content gaps** — missing topics that competitors cover
+5. **Orphan pages** — add internal links
+6. **Technical issues** — broken links, slow load times, missing meta tags
+7. **Thin content** — pages below 300 words on competitive topics
+
+### Step 7: Fix and Re-Audit
+
+**Via MCP:**
+```
+1. For content issues: `start_optimization` → `get_optimization` → `apply_optimization`
+2. For missing content: `create_brief` → `generate_content` → optimize → publish
+3. Re-run audit: `start_audit` again and compare scores
+```
+
+**Via CLI:**
+```bash
+frase optimize start --content-id <id>
+frase optimize apply <opt-id> --suggestion <sug-id>
+frase audit start --site-id <id>        # re-audit after fixes
+```
+
+## Interpreting Scores
+
+### SEO Score
+- **70+:** Well-optimized for search engines
+- **50-69:** Missing some important optimization elements
+- **< 50:** Needs substantial optimization work
+
+### GEO Score
+- **70+:** Well-structured for AI citation
+- **50-69:** Some AI-readiness but gaps exist
+- **< 50:** Not structured for AI search engines
+
+### Health Score
+- **80+:** Healthy site — minor maintenance needed
+- **60-79:** Some issues need attention
+- **< 60:** Significant issues impacting performance
+
+## Key Principles
+
+1. **Health check first** — get the overview before deep-diving
+2. **Fix cannibalization early** — it's a silent traffic killer
+3. **Content before technical** — content fixes have higher traffic impact
+4. **Prioritize by traffic impact** — fix high-traffic pages first
+5. **Re-audit after fixes** — verify improvements before moving on
+6. **Regular audits** — run monthly to catch issues early

package/package.json

@@ -0,0 +1,44 @@
+{
+  "name": "@frase/agent-skills",
+  "version": "0.3.0",
+  "description": "Agent skills for Frase — teach AI assistants how to research, write, optimize, host, and monitor content for SEO, GEO, and AI search visibility. 9 skills covering the full content lifecycle including Frase CMS hosting with auto-optimization. Works with Claude Code, Cursor, Windsurf, Lovable, Replit, and any MCP-compatible tool.",
+  "keywords": [
+    "frase",
+    "agent-skills",
+    "mcp",
+    "seo",
+    "geo",
+    "aeo",
+    "ai-visibility",
+    "content-automation",
+    "claude-code",
+    "cursor",
+    "windsurf",
+    "lovable",
+    "replit",
+    "ai-agent"
+  ],
+  "license": "MIT",
+  "author": {
+    "name": "Frase",
+    "email": "hello@frase.io",
+    "url": "https://frase.io"
+  },
+  "homepage": "https://frase.io/agents",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/frase-io/newfrase.git",
+    "directory": "packages/agent-skills"
+  },
+  "bugs": {
+    "url": "https://github.com/frase-io/newfrase/issues"
+  },
+  "files": [
+    "frase-*/SKILL.md",
+    "skills.json",
+    "README.md"
+  ],
+  "publishConfig": {
+    "access": "public"
+  }
+}

package/skills.json

@@ -0,0 +1,77 @@
+{
+  "name": "@frase/agent-skills",
+  "version": "0.3.0",
+  "description": "Agent skills for Frase SEO, GEO, AEO & AI Visibility platform — teaches AI assistants how to use Frase tools via MCP or CLI",
+  "author": "Frase <hello@frase.io>",
+  "homepage": "https://frase.io",
+  "repository": "https://github.com/frase-io/newfrase",
+  "skills": [
+    {
+      "name": "frase-content-pipeline",
+      "file": "frase-content-pipeline/SKILL.md",
+      "description": "End-to-end content workflow: research → brief → write → optimize → publish → monitor. Works via MCP or CLI.",
+      "tags": ["seo", "geo", "content", "publishing", "optimization"]
+    },
+    {
+      "name": "frase-seo-audit",
+      "file": "frase-seo-audit/SKILL.md",
+      "description": "Run and interpret SEO/GEO audits — site crawls, scoring, site health, cannibalization detection, and prioritized fix workflows. Works via MCP or CLI.",
+      "tags": ["seo", "geo", "audit", "site-health", "cannibalization"]
+    },
+    {
+      "name": "frase-ai-visibility",
+      "file": "frase-ai-visibility/SKILL.md",
+      "description": "Track and improve brand visibility across 8 AI search engines (ChatGPT, Perplexity, Claude, Google AI Overviews, Copilot, Grok, DeepSeek, Gemini). Works via MCP or CLI.",
+      "tags": ["geo", "aeo", "ai-search", "visibility", "monitoring"]
+    },
+    {
+      "name": "frase-content-optimization",
+      "file": "frase-content-optimization/SKILL.md",
+      "description": "Score, optimize, and improve content for SEO and GEO performance. Works via MCP or CLI.",
+      "tags": ["seo", "geo", "optimization", "content-scoring"]
+    },
+    {
+      "name": "frase-geo-optimization",
+      "file": "frase-geo-optimization/SKILL.md",
+      "description": "Optimize content for AI engine citation (GEO/AEO) — quotability scoring, content structuring, cross-platform citation analysis, and gap diagnosis. Works via MCP or CLI.",
+      "tags": ["geo", "aeo", "ai-citation", "quotability", "optimization"]
+    },
+    {
+      "name": "frase-atomization",
+      "file": "frase-atomization/SKILL.md",
+      "description": "Break long-form content into atomic pieces (social posts, email snippets, ad copy, threads) — batch atomize and export across channels. Works via MCP or CLI.",
+      "tags": ["atomization", "content-repurposing", "social-media", "distribution"]
+    },
+    {
+      "name": "frase-playbook-automation",
+      "file": "frase-playbook-automation/SKILL.md",
+      "description": "Automate multi-step content workflows using playbooks, templates, and content rules — chain research to publish with quality gates. Works via MCP or CLI.",
+      "tags": ["automation", "playbooks", "templates", "rules", "governance"]
+    },
+    {
+      "name": "frase-competitive-intelligence",
+      "file": "frase-competitive-intelligence/SKILL.md",
+      "description": "Deep competitive analysis, opportunity detection, and topic clustering — analyze competitors, score opportunities, organize content into clusters. Works via MCP or CLI.",
+      "tags": ["competitive-analysis", "opportunities", "clusters", "gap-analysis"]
+    },
+    {
+      "name": "frase-cms-publishing",
+      "file": "frase-cms-publishing/SKILL.md",
+      "description": "Multi-platform publishing — FraseCMS, WordPress, Webflow, Sanity, scheduling, auto-optimization, and content versioning. Works via MCP or CLI.",
+      "tags": ["publishing", "cms", "wordpress", "webflow", "sanity", "auto-optimization"]
+    }
+  ],
+  "transports": {
+    "mcp_server": {
+      "package": "@frase/mcp-server",
+      "install": "claude mcp add frase -e FRASE_API_KEY -- npx -y @frase/mcp-server",
+      "required_env": ["FRASE_API_KEY"]
+    },
+    "cli": {
+      "package": "@frase/cli",
+      "install": "npm install -g @frase/cli",
+      "npx": "npx @frase/cli",
+      "required_env": ["FRASE_API_KEY"]
+    }
+  }
+}