HOL Guard rule for `approve_plugin`?

Abierto
#507 0 comentarios 0 reacciones 0 asignados Ver en GitHub

Los mantenedores suelen responder en 1 día

Nadie ha tomado este issue todavía.

Evaluación

Dificultad
4/5
Tiempo estimado
3-5 días
Aptitud para principiantes
56/100
Tipo de issue
Nueva funcionalidad
Claridad
Bien especificado
Estado de actividad
Activo
Stack tecnológico
github, typescript

Línea de trabajo

Read the upstream community-plugins behavior and the HOL Guard MCP contribution guide and matching Builder example linked in the issue. Export the complete tools/list inventory, generate and review the profile for approve_plugin, and retain existing handling for get_plugin and unknown tools. Validate and preview the kit, run the named MCP contribution and trust checks, then submit a draft PR with the profile, behavior cases, and validation results.

Escrito por el modelo de indexación a partir del texto del issue.

Descripción

Hey, we maintain HOL Guard. For people using community-plugins through an AI agent, an optional Guard profile would ask for human approval before the approve_plugin action.

Would you contribute a profile covering approve_plugin? The first draft can focus on that operation and get_plugin as the comparison case. The Builder handoff below is specific to this project; you generate and review the files, then open your PR. A draft with the reviewed profile and validation results is enough to start review. Local Guard protection is free and needs no account. The extension stays opt-in, and your repo needs no code or CI changes.

Builder handoff for this project

Scope: cursor/community-plugins, approve_plugin, and the get_plugin comparison. Upstream reference.

Existing safeguard documented by the project:

Write tools are rate-limited per admin (mcp-moderation Vercel Firewall rule), and every change from the web admin or MCP is recorded in the append-only plugin_moderation_events table.

You can give these instructions to your coding agent:

  1. Work in your own fork/branch of hashgraph-online/hol-guard. Export the complete MCP tools/list inventory using your project's documented workflow, including every page. Confirm the package/launcher or HTTPS identity and version from your own documentation; do not guess them from the repository name.
  2. Follow the matching Builder example to generate a review kit from that saved export. Use physical paths without symlink components and a new output directory whose parent already exists. The Builder reads the export offline; it does not run your tool.
  3. Propose review for approve_plugin. Use get_plugin as the safe comparison case, retaining its existing handling. Unknown tools inherit existing policy. Do not add an allow default based only on tool annotations or a remote server name. Review the generated decisions against the upstream behavior; record the rationale and public source references in a copy of review.json and set reviewed=true for the decisions you reviewed. Preserve the discoveryDigest, operation IDs, other entries, and supported riskClasses. Regenerate with extensions generate --from snapshot, using the original discovery.json and your explicit review file.
  4. Run hol-guard extensions validate on that reviewed kit. Preview hol-guard extensions apply against your fork, inspect the file plan, and apply only with its printed --expected-plan digest. Keep the MCP JSON contract under contributions/mcp-servers/ and the external trust binding. Run the MCP contribution test and trust checks named in the MCP guide. Record the risky and comparison-tool decisions in your PR. The MCP Builder does not emit a command fixture; do not invent a command-source document or a Python detector. Follow the contribution checks for this contribution type.
  5. Review the full diff and behavior cases, then open your own draft PR with the contribution files and validation results. Explain what needs approval and what retains its existing handling. Keep the contribution External and off until locally enabled. Do not add secrets, change your source repository's CI, or enable Guard protections as part of generating the contribution.

A draft containing the profile/source and meaningful behavior cases is enough to begin review. Report a failed Builder command with its error before working around it.

Lenguaje dominante
TypeScript
Estrellas
4k
Forks
674
Merge medio
8 min
PR fusionados (30 d)
28

Preparar el entorno

Este proyecto no incluye contenedor de desarrollo, Dockerfile ni guía de contribución, así que la configuración corre por tu cuenta: empieza por su README y consulta nuestra guía para la primera contribución para los pasos generales.

Primeros pasos

  1. Lee el issue completo y luego la guía de contribución del proyecto.
  2. Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
  3. Haz un fork del repositorio y trabaja en una rama.
  4. Abre un pull request que haga referencia al número del issue.

Más de cursor/community-plugins

Todos los issues de cursor/community-plugins

Issues similares

Más issues de TypeScript

Recibe los nuevos issues en tu correo

Un resumen breve de issues de GitHub para principiantes.