Features

Anonymous one-shot site audit — security headers, SEO basics, performance, accessibility, best-practices. No signup. Returns a public share URL + badge.

POST /api/v1/sync/onboard with {url, name, email}. Returns api_key + site_id. Already-registered URLs return the existing key — no duplicate sites, no manual recovery.

POST /api/v1/sync/bulk with ARP §3 reports. Severity, summary, metrics, warnings (with fingerprint + lifecycle), recommendations. A-F data quality grade in the response.

POST /api/v1/connect returns critical / warning / recommendations buckets filtered by your stack, plus cross_site_tips ranked by Beta-Bernoulli confidence × applicability.

GET /api/v1/sync/personalized-rules — Tier-2 patterns from peer sites that solved related issues, ranked per your tech_stack, site_type, and history.

GET /api/v1/sync/network-position — your score and grade vs the network p50 / avg, with percentile and grade distribution.

Patterns are confidence-weighted, fingerprinted, and tracked through active → solved → resurfaced lifecycle. Status filtering normative on consumers.

POST /api/v1/sync/analyze-actions runs a fresh scan + LLM synthesis and returns ARP §3.4-shape recommendations with title / why / code_snippet / provenance / confidence. 5-minute response cache so UI re-fires don't re-pay the LLM round-trip.

openinference-instrumentation-agentminds on PyPI. Apache-2.0. Auto-instruments push + connect calls with OpenInference attribute namespace.

JSON-RPC 2.0 over Streamable HTTP at /mcp/v1/. Tool catalog (push / connect / status / query_pattern) currently flagged experimental — spec-handshake first, direct delegation in MCP server v1.3.

/.well-known/ entries for ARP, A2A AgentCard, MCP server, OASF descriptor, plus OpenInference SDK. Discoverable without out-of-band coordination.

Sentry envelope ingest with fingerprint normalization, OTel GenAI translator, Langfuse / LangSmith bridge primitives.

W3C Verifiable Credentials binding for cross-org agent delegation. Cryptographic identity proof via AGNTCY OASF Identity Service.

Five skills published in agentskills.io frontmatter format: agentminds-connect, agentminds-status, agentminds-agents, agentminds-pool, agentminds-push-report.

Sidebar that surfaces your site's open warnings, recommendations, and recently observed patterns in-IDE. Same auth as the REST API.

agentminds-cli on npm and pip — onboard, push reports, pull recommendations, query patterns from a terminal.