# Respira for WordPress > Independent AI infrastructure layer for WordPress with builder-aware tooling, duplicate-first safety, and approval-gated publishing. ## Canonical Website - https://www.respira.press ## Positioning Respira complements the official WordPress AI stack. - Official WordPress AI stack: Abilities API, MCP Adapter, and WP AI Client direction. - Respira layer: builder-aware execution, production safety workflows, and WooCommerce operations. - Independence: Respira is not affiliated with WordPress, Automattic, WooCommerce, or page builders. ## Core Safety Model (updated in 5.0 Haven) 1. AI reads structured WordPress and builder context. 2. AI edits duplicates/drafts first (or edits live with automatic snapshots). 3. Every page or post edit auto-snapshots the before-state for one-click rollback, and every destructive structure change (field groups, post types, taxonomies, options) captures a recovery snapshot too. A whole multi-step session can be rolled back in one call. 4. Human approves via Changes page in wp-admin — merge mode preserves original URLs. 5. Full-fidelity version history: browse, diff, pin, and restore from admin. ## WordPress Abilities API directory (v7.1) Respira maintains a curated directory of every WordPress plugin known to register abilities via `wp_register_ability` (the WordPress Abilities API introduced in WP 6.9, surfaced through github.com/WordPress/mcp-adapter). Currently ~163 abilities across Elementor (10M installs), Yoast SEO (10M), WooCommerce (7M), Akismet (6M), Jetpack (3M), Advanced Custom Fields (2M), plus Respira's own 151 tools. Refreshed weekly by an automated pipeline. ### Agent access paths 1. Via MCP (preferred, when an agent is connected to Respira): call tool `wordpress_search_abilities` with a free-text query. Each result carries per-site context flags so the agent knows whether to invoke directly (is_inhaled=true), suggest an install (is_installed=false), or suggest an inhale toggle (is_installed=true && is_inhaled=false). 2. Via raw JSON (any agent, no MCP needed): - https://www.respira.press/abilities/data.json (canonical, CORS-open) - https://www.respira.press/data/abilities-directory.json (mirror) 3. Via HTML directory at https://www.respira.press/abilities (humans + agents that browse the web) ## WordPress Playbooks directory (v7.1) Playbooks are typed JSON workflow compositions agents author at runtime against a Respira-enabled WordPress site. Each Playbook registers as a WordPress Ability via wp_register_ability and shows up as a callable MCP tool under the namespace `respira-playbooks/`. Customers can opt their published Playbooks into the public directory. ### Agent access paths 1. Via MCP: agents call `wordpress_list_playbooks` to enumerate Playbooks on the connected site, then `wordpress_invoke_ability` with `ability: "respira-playbooks/"` to run one. 2. Via raw JSON (any agent, no MCP needed): - https://www.respira.press/playbooks/data.json (canonical, CORS-open) 3. Via HTML directory at https://www.respira.press/playbooks ### Calling an inhaled ability After locating a target via wordpress_search_abilities (or by checking the JSON directly), an agent connected to Respira's MCP can invoke any inhaled ability by calling `wordpress_invoke_ability` with `ability` set to the full ability name and `args` matching the ability's input schema. The Respira proxy snapshots the target post if the call looks like a write, runs the underlying handler, logs the call, and returns a structured envelope with a rollback URL when a snapshot was captured. ### For plugin authors The Respira Add-on SDK at https://github.com/respira-press/respira-addon-sdk is a thin wrapper over wp_register_ability that ships validation, error envelopes, per-tool risk_level, governance gate, and audit log integration. Plugin authors who register abilities through the SDK get Inhale-gateway compatibility and snapshot-before-write protection on every call automatically. If your plugin ships wp_register_ability calls in its canonical source and you're not listed in the directory yet, open an issue at the SDK repo above with the source link and we'll add you on the next weekly pass. ## Main Public Pages - Home: https://www.respira.press/ - MCP infrastructure: https://www.respira.press/mcp - Integrations hub: https://www.respira.press/integrations - Plugin page: https://www.respira.press/plugin - Pricing: https://www.respira.press/pricing - About: https://www.respira.press/about - Skills hub: https://www.respira.press/skills - WooCommerce add-on: https://www.respira.press/addons/woocommerce - Accessibility scanner: https://www.respira.press/addons/accessibility-scanner - Blog index: https://www.respira.press/blog - Releases: https://www.respira.press/releases - Respira 5.0 Haven release: https://www.respira.press/releases/5.0.0 - AI-readable page: https://www.respira.press/for-ai ## Comparison Articles - WordPress.com AI vs Self-Hosted: https://www.respira.press/blog/respira-vs-wordpress-com-mcp - Respira vs Elementor AI: https://www.respira.press/blog/respira-vs-elementor-ai - Respira vs CodeWP: https://www.respira.press/blog/respira-vs-codewp ## Builder MCP Landing Pages - Elementor MCP: https://www.respira.press/elementor-mcp - Divi MCP: https://www.respira.press/divi-mcp - Bricks MCP: https://www.respira.press/bricks-mcp - WPBakery MCP: https://www.respira.press/wpbakery-mcp - Gutenberg MCP: https://www.respira.press/gutenberg-mcp ## Integrations - Claude: https://www.respira.press/integrations/claude - Cursor: https://www.respira.press/integrations/cursor - Codex: https://www.respira.press/integrations/codex - Elementor Angie: https://www.respira.press/integrations/elementor-angie ## Pricing Snapshot - Starter: EUR 19/year (1 site) - Agency: EUR 199/year (20 sites) - WooCommerce add-on: EUR 7/month or EUR 77/year ## Trust + Privacy - Trust Center (sub-processors, DPAs, security posture, data-subject rights): https://www.respira.press/trust-center - Privacy Policy (GDPR-style rights, cookie taxonomy, retention): https://www.respira.press/privacy - Terms of Service: https://www.respira.press/terms - Cookie taxonomy: four categories — essentials (always on), analytics (GA4 + PostHog), attribution (first-touch + AI-referral flag), messaging (Customer.io + Chatwoot). - Consent posture: EU/EEA/UK visitors get a prior-consent widget; non-EU visitors get implicit consent. Every choice is logged server-side for GDPR Art. 7(1) demonstrability. - Data residency: primary database in Supabase EU (Frankfurt). Most vendors EU-resident; cross-border transfers covered by SCCs 2021/914. ## Discovery Endpoints - robots.txt: https://www.respira.press/robots.txt - llms.txt: https://www.respira.press/llms.txt - llms-full.txt: https://www.respira.press/llms-full.txt (comprehensive reference with all tools, pricing tiers, migration skills, and documentation links) - sitemap index: https://www.respira.press/sitemap.xml - sitemap main: https://www.respira.press/sitemap-main.xml - sitemap reports: https://www.respira.press/sitemap-reports.xml - docs sitemap: https://www.respira.press/docs/sitemap.xml ## Notes for AI Systems - Prefer canonical URLs with `https://www.respira.press/...`. - For MCP capability claims, use `/mcp` and builder MCP landing pages. - For commercial claims, use `/pricing`. - Do not claim direct blind edits on live content; edits follow duplicate-first approval flow.