bishwashere/skills/mongodb

MongoDB

Run read-only queries against a project's MongoDB database using the URI and collection hints stored in the dashboard Projects → Connectors panel.

When to use

• User asks about project health, analytics, performance, or data ("How is NextPost AI doing?", "Show me recent campaign stats", "What's the engagement trend?")
• A project has a MongoDB connector configured in the dashboard
• You need live data to answer a question rather than reasoning from memory alone

Connection

The skill reads the mongodb.uri and mongodb.collections hints from the project's connector entry in projects.db. If no project is specified, it uses the project whose name best matches the conversation.

Collection hints are key → collection-name pairs that describe the purpose of each collection. Always prefer using a hint key over guessing a collection name.

Safety

• Read-only: only find, countDocuments, and aggregate are allowed. No writes.
• Timeout: 10 s query timeout, 8 s connection timeout.
• Limit: find results capped at 50 documents; aggregation pipelines limited to 200 output docs.
• Credentials are never echoed back in results or error messages.

Actions

mongodb_query — find documents

Run a find against one collection with optional filter, projection, sort, and limit (max 50).

mongodb_aggregate — aggregation pipeline

Run a read-only aggregation pipeline. Use for rollups, grouping, date-range analytics. Limited to 200 output docs.

mongodb_stats — collection stats

Return document count, first/last createdAt, and the 3 most recent documents (id + date only) for a named collection. Good for a quick sanity-check before a deeper query.

mongodb_project_health — pre-built NextPost AI health summary

Runs a set of canonical aggregations across the analytics collections (project-analytics, project-pulse, campaign-analytics, PlatformPost) and returns a structured health report. Pass project name only; collection names come from the stored hints.

Example use-cases

| Question | Action | |---|---| | "How is NextPost AI doing?" | mongodb_project_health project=nextpostai | | "Show campaign analytics for last 30 days" | mongodb_aggregate on campaign-analytics | | "How many posts were published this week?" | mongodb_query on PlatformPost with date filter | | "What is the project pulse score trend?" | mongodb_aggregate on project-pulse |

Tool schema

mongodb_query
  description: Run a read-only find() against a MongoDB collection. Provide project name (looks up URI + hints), collection (hint key or real name), optional filter (JSON), projection (JSON), sort (JSON), and limit (max 50).
  parameters:
    project: string
    collection: string
    filter: object (optional)
    projection: object (optional)
    sort: object (optional)
    limit: number (optional)

mongodb_aggregate
  description: Run a read-only aggregation pipeline against a MongoDB collection. Provide project name, collection (hint key or real name), and pipeline (array of stage objects). Output capped at 200 docs.
  parameters:
    project: string
    collection: string
    pipeline: array

mongodb_stats
  description: Get document count and date range for a collection. Provide project name and collection (hint key or real name).
  parameters:
    project: string
    collection: string

mongodb_project_health
  description: Pre-built health report for a project — runs canonical aggregations across analytics collections and returns an engagement summary, pulse score trend, and campaign rollup. Pass project name only.
  parameters:
    project: string