convez.ai
MCP ServerDevOps & InfraestruturaAvançado
osinfra-io

Techne MCP Server

osinfra-io/pt-techne-mcp-server

MCP server para renderização tipada de infraestrutura e configurações de plataforma

Ver no GitHub
BomPluginScore · de 0 a 100

Servidor MCP que fornece contexto de plataforma e ferramentas tipadas para assistentes de IA, especializado em renderização determinística de configurações HCL e artefatos de infraestrutura. Integra-se com repositórios GitHub da osinfra para exposição de ferramentas testadas que evitam escrita manual de código de infraestrutura.

Instalação

$npx convez-ai osinfra-io/pt-techne-mcp-server

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
0
Acompanhando
0
Saúde
87%
Último commit
há 3 dias
Versão
v0.7.0
Issues abertas
2
Contribuidores
3

Casos de uso

  • Geração determinística de configurações HCL para infraestrutura como código
  • Validação e renderização de especificações de equipe via ferramentas MCP
  • Integração de IA com fluxos de infraestrutura através de ferramentas tipadas

Veredito

MCP server robusto e bem-mantido, voltado para automação de infraestrutura em ambiente osinfra. Requer tokens GitHub e conhecimento técnico de DevOps/IaC para máxima utilidade; deploy simples via Docker ou binário pré-compilado com verificação criptográfica Sigstore.

README

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

Techne MCP Server

Tests Dependabot

Model Context Protocol (MCP) server providing platform context and tools to AI assistants. Exposes deterministic, typed tools so platform agents call a tested renderer instead of writing HCL by hand.

Usage

Docker (recommended)

Add this entry to your MCP client config (e.g. .copilot/mcp.json, mcp.json):

{
  "mcpServers": {
    "pt-techne-mcp-server": {
      "command": "docker",
      "args": [
        "run", "-i", "--rm",
        "-e", "GITHUB_TOKEN",
        "ghcr.io/osinfra-io/pt-techne-mcp-server:latest"
      ]
    }
  }
}

Download binary

Pre-built binaries for Linux (amd64, arm64) are attached to each GitHub release:

# Download the latest release for your architecture
gh release download --repo osinfra-io/pt-techne-mcp-server --pattern '*linux-amd64'
chmod +x pt-techne-mcp-server-linux-amd64

Then configure your MCP client:

{
  "mcpServers": {
    "pt-techne-mcp-server": {
      "command": "/path/to/pt-techne-mcp-server-linux-amd64",
      "env": {
        "GITHUB_TOKEN": "<YOUR_TOKEN>"
      }
    }
  }
}

Build from source

go install github.com/osinfra-io/pt-techne-mcp-server/cmd/pt-techne-mcp-server@latest
{
  "mcpServers": {
    "pt-techne-mcp-server": {
      "command": "pt-techne-mcp-server",
      "env": {
        "GITHUB_TOKEN": "<YOUR_TOKEN>"
      }
    }
  }
}

Verification

Release artifacts are signed with cosign keyless (Sigstore OIDC). Install cosign to verify.

Container image:

cosign verify ghcr.io/osinfra-io/pt-techne-mcp-server:<tag> \
  --certificate-identity-regexp='https://github.com/osinfra-io/pt-techne-mcp-server/.github/workflows/release.yml@refs/tags/.*' \
  --certificate-oidc-issuer='https://token.actions.githubusercontent.com'

Binary:

gh release download --repo osinfra-io/pt-techne-mcp-server --pattern '*linux-amd64*'

cosign verify-blob pt-techne-mcp-server-linux-amd64 \
  --bundle pt-techne-mcp-server-linux-amd64.bundle \
  --certificate-identity-regexp='https://github.com/osinfra-io/pt-techne-mcp-server/.github/workflows/release.yml@refs/tags/.*' \
  --certificate-oidc-issuer='https://token.actions.githubusercontent.com'

Configuration

Set GITHUB_TOKEN with access to osinfra-io/pt-logos, osinfra-io/pt-corpus, osinfra-io/pt-pneuma, and osinfra-io/pt-ekklesia-docs. Without it, GitHub-backed tools return not_configured; offline tools (validate_team_spec, render_team_tfvars, render_team_docs_index, render_sidebar_patch) still work.

For local development, gh auth token is the simplest option. See docs/configuration.md for full details on token scopes, sources, and operational error codes.

Tools

ToolDescription
validate_team_specValidate a team spec against the JSON Schema
render_team_tfvarsRender a spec to canonical pt-logos .tfvars bytes
open_team_prValidate, render, and open/update a PR on pt-logos
list_teamsList all teams from pt-logos@main
get_teamGet a single team's spec and docs pages
lookup_userFind every team/role where a user appears
find_repoFind which team(s) own a repository
next_available_cidrsCompute the next N unallocated GKE subnet CIDR slots
render_corpus_helpersInsert a workspace into pt-corpus helpers.tofu
render_pneuma_helpersInsert a workspace into pt-pneuma helpers.tofu
render_team_docs_indexRender a team's docs index page
render_sidebar_patchPatch pt-ekklesia-docs sidebars.js with a team entry
open_team_docs_prRender docs + sidebar and open/update a PR on pt-ekklesia-docs

All spec-accepting tools handle both JSON objects and JSON-encoded strings (for LLM double-encoding quirks). Write tools are idempotent — identical input returns action: "noop".

Local development

Requires Go 1.26.3.

go build ./cmd/pt-techne-mcp-server            # build the binary
go test -race ./...                             # run tests
golangci-lint run                               # lint (uses .golangci.yml)

Schema maintenance (run after editing internal/spec/schema_embed.json):

go run ./internal/schemadoc schema/team.schema.json docs/schema.md  # regenerate docs

Documentation

DocumentDescription
docs/contributing.mdRepo layout, architecture, how to add a tool
docs/configuration.mdToken setup, operational error codes, tool semantics
docs/schema.mdGenerated reference for every team spec field
schema/team.schema.jsonTeam spec schema (symlink to internal/spec/schema_embed.json)

License

GPL-2.0. See LICENSE.

Ver o original no GitHub

Bom para

Relacionados