convez.ai
MCP ServerDevOps & InfraestruturaIntermediário
frozo-ai

EdgeGate MCP — Gates de Regressão IA em Edge

frozo-ai/edgegate-mcp

Configure gates de regressão IA em hardware Snapdragon via prompt

Ver no GitHub
InicialPluginScore · de 0 a 100

Servidor MCP que expõe a plataforma EdgeGate como 30+ ferramentas para configurar e executar testes de regressão de modelos IA em dispositivos Snapdragon (Galaxy S24, S23, etc.) diretamente do Claude Code, Cursor ou Claude Desktop. Gera bundles de evidência assinados para gates de CI/CD com métricas de latência, memória e desempenho.

Instalação

$npx convez-ai frozo-ai/edgegate-mcp

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
1
Forks
1
Acompanhando
0
Saúde
50%
Último commit
há 1 mês
Versão
v0.7.0
Issues abertas
1
Contribuidores
2

Casos de uso

  • Validar desempenho de modelos ONNX em dispositivos reais antes de deploy
  • Integrar testes de regressão em CI/CD (GitHub Actions)
  • Comparar runs e detectar degradação de performance entre versões
  • Gerenciar modelos HuggingFace e Qualcomm AI Hub de forma integrada

Veredito

Ferramenta especializada e bem-integrada para quem trabalha com ML ops em edge computing. Reduz fricção ao automatizar testes em Snapdragon via prompt, mas requer acesso à plataforma EdgeGate e dispositivos reais para valor completo.

README

A tradução para o português deste README ainda está na fila.

edgegate-mcp

MCP server for EdgeGate — set up edge-AI regression gates on Snapdragon devices directly from Claude Code, Cursor, or Claude Desktop.

What does it do?

EdgeGate runs AI model regression tests on real Snapdragon hardware via Qualcomm AI Hub, then produces signed evidence bundles you can attach to CI gates. This npm package exposes EdgeGate's REST API as 30+ MCP tools, plus bundled skills, so you can drive the whole flow from a prompt:

> Use the edgegate MCP to set up a CI gate for my MobileNet ONNX model.
> Gates: inference_time_ms ≤ 10, peak_memory_mb ≤ 150.
> Devices: Galaxy S24, Galaxy S23.

Install

# 1. Generate an API key in the EdgeGate dashboard
# https://edgegate.frozo.ai/workspace/<id>/settings#api-keys

# 2. Run the installer (writes config for Claude Code / Cursor / Desktop)
npx edgegate-mcp-install

Restart your MCP client. Done.

Manual config

If you'd rather edit config files yourself, the server is a standard stdio MCP. Add this to your client's config:

Claude Code (~/.claude.json)

{
  "mcpServers": {
    "edgegate": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "edgegate-mcp"],
      "env": {
        "EDGEGATE_API_KEY": "egk_live_...",
        "EDGEGATE_API_URL": "https://edgegateapi.frozo.ai"
      }
    }
  }
}

Cursor (~/.cursor/mcp.json)

Same shape as Claude Code without the type field.

Claude Desktop (~/Library/Application Support/Claude/claude_desktop_config.json on macOS)

Same shape as Cursor.

Tools

See docs/tools.md for the full tool reference. Quick list:

ToolPurpose
edgegate_setup_workspacePick / confirm the active workspace
edgegate_create_pipelineDefine a new regression pipeline
edgegate_run_gateTrigger a run
edgegate_check_statusPoll a run for status + metrics
edgegate_get_reportList recent runs
edgegate_get_audit_reportFetch the signed audit PDF
edgegate_setup_github_actionGenerate the GitHub Actions workflow + secret commands
edgegate_compare_runsDiff two runs — gate flips, metric deltas, per-device breakdown, REGRESSION / IMPROVEMENT / NEUTRAL verdict
edgegate_export_run_reportSave a complete run report as a markdown file to disk (returns the file path + preview)
edgegate_import_huggingface_modelImport a HuggingFace ONNX model — returns artifact_id ready for edgegate_create_pipeline. Uses the workspace's personal HF token when connected; otherwise anonymous
edgegate_connect_huggingfaceStore a personal HuggingFace token in the workspace so imports can read private / gated / Qualcomm-org repos. Validated against HF whoami before encryption
edgegate_get_huggingface_integrationShow whether a HF token is currently connected (no plaintext)
edgegate_disconnect_huggingfaceDelete the HF integration; future imports fall back to anonymous access
edgegate_connect_qaihubConnect / rotate a Qualcomm AI Hub token (encrypted at rest, never returned in plaintext)
edgegate_get_qaihub_integrationShow the connection status of the workspace's AI Hub token
edgegate_disconnect_qaihubPermanently delete the AI Hub integration
edgegate_create_workspaceCreate a new workspace (caller becomes owner)
edgegate_list_api_keysList API keys in the workspace (no plaintext)
edgegate_create_api_keyMint a new API key — plaintext returned exactly once
edgegate_revoke_api_keyRevoke an API key by id (audit row preserved)
edgegate_list_membersList members + roles
edgegate_invite_memberAdd an existing EdgeGate user to the workspace at a given role
edgegate_change_member_roleChange a member's role (cannot downgrade the last owner)
edgegate_remove_memberRemove a member (their pipelines/runs are preserved)
edgegate_list_promptpacksList all promptpacks in a workspace (id, version, case count, published status)
edgegate_create_promptpackCreate a new promptpack with test cases (prompts, expected outputs, per-case overrides)
edgegate_publish_promptpackPublish a promptpack version so it is usable in pipelines (completes the create → publish → use lifecycle)
edgegate_register_byo_bucketEnterprise. Register the workspace's own S3 bucket + IAM role as a BYO storage grant. Returns the External ID to paste into the role's trust policy
edgegate_check_byo_bucketEnterprise. Re-run the AssumeRole + HeadObject readiness probe; renders the typed BYO_* error code with a fix checklist
edgegate_register_byo_artifactEnterprise. Register an existing s3://<bucket>/<key> URI as an EdgeGate Artifact (no bytes uploaded — HeadObject only). Returns an artifact_id ready for edgegate_create_pipeline
edgegate_disconnect_byo_bucketEnterprise. Delete the BYO grant. Refuses (409) if artifacts still reference it
edgegate_get_byo_auditEnterprise. Paginated append-only audit log of every AssumeRole / HeadObject / GetObject / KMS Decrypt call. aws_request_id joins to your own CloudTrail

Skills

Slash commands you can invoke directly:

  • /edgegate-init — full onboarding flow (zero → CI gate)
  • /edgegate-gate — trigger a run on an existing pipeline
  • /edgegate-status — check a run's status + metrics
  • /edgegate-audit — fetch the evidence bundle for a run
  • /edgegate-compare — diff two runs (auto-baseline) with REGRESSION/IMPROVEMENT/NEUTRAL verdict
  • /edgegate-export — save a run report as a markdown file (for PR comments, Slack, compliance)
  • /edgegate-import — import a Hugging Face ONNX model (anonymous, or workspace-token for private / gated / qualcomm-org repos)
  • /edgegate-connect-huggingface — attach a personal HuggingFace token to the workspace so the import flow can read private / gated / qualcomm-org repos
  • /edgegate-connect-qaihub — connect a Qualcomm AI Hub token so runs can actually compile and profile on Snapdragon devices
  • /edgegate-workspace-setup — bootstrap a brand-new workspace end-to-end (create → connect AI Hub → API key → invite teammates)
  • /edgegate-members — list / invite / change role / remove workspace members
  • /edgegate-promptpacks — list existing promptpacks or create a new one with generated test cases
  • /edgegate-byo-storage — Enterprise BYO storage onboarding (register grant → paste External ID → verify probe → first artifact → first run → audit log)

License

MIT — see LICENSE.

Ver o original no GitHub

Bom para

Relacionados