Agent Skills › opskat/opskat › mongodb

mongodb

GitHub

提供 MongoDB 集合的查询与修改能力，支持 find、insert、update、delete 及聚合等操作。通过 exec 执行命令，使用 JSON 语法传递查询参数，适用于数据检索、写入、更新及批量处理场景。

internal/ai/skills/mongodb/SKILL.md opskat/opskat

Trigger Scenarios

需要查询 MongoDB 数据需要向 MongoDB 插入或更新文档需要从 MongoDB 删除数据需要对 MongoDB 数据进行聚合分析

Install

npx skills add opskat/opskat --skill mongodb -g -y

More Options

Non-standard path

npx skills add https://github.com/opskat/opskat/tree/main/internal/ai/skills/mongodb -g -y

Use without installing

npx skills use opskat/opskat@mongodb

指定 Agent (Claude Code)

npx skills add opskat/opskat --skill mongodb -a claude-code -g -y

安装 repo 全部 skill

npx skills add opskat/opskat --all -g -y

预览 repo 内 skill

npx skills add opskat/opskat --list

SKILL.md

Frontmatter

{
    "name": "mongodb",
    "description": "Query and modify MongoDB collections via exec, using an operation + collection + JSON query syntax."
}

MongoDB assets

Command syntax

<operation> [collection] [--db=<database>] [--query=<json>]

find users --query='{"filter":{"age":{"$gt":21}},"limit":10}'
findOne users --query='{"filter":{"_id":"abc"}}'
aggregate events --query='{"pipeline":[{"$match":{"type":"click"}}]}'
countDocuments users
insertOne users --query='{"document":{"name":"alice"}}'
updateMany users --query='{"filter":{"active":false},"update":{"$set":{"archived":true}}}'
deleteMany logs --query='{"filter":{"level":"debug"}}'
listDatabases
listCollections --db=analytics

Supported operations: find, findOne, insertOne, insertMany, updateOne, updateMany, deleteOne, deleteMany, aggregate, countDocuments, listDatabases, listCollections. Anything else is rejected.

Query sub-keys

--query is a single JSON object. Which sub-keys apply depends on the operation:

find: filter, sort, projection, limit, skip
findOne: filter, projection (sort / limit / skip are silently ignored)
insertOne: document · insertMany: documents
updateOne / updateMany: filter, update
deleteOne / deleteMany: filter
aggregate: pipeline
countDocuments: filter

Notes

Always single-quote --query — it is JSON and contains spaces and braces.
find returns at most 100 documents when limit is omitted. Pass an explicit limit when you need to know you have the whole result set.
Every operation except listDatabases needs a database. --db names it; if you omit --db, the asset's configured default database is used. When the asset has no default database configured, omitting --db fails — pass it explicitly.
listDatabases and listCollections take no collection; every other operation requires one.
Omitting --query means "no filter": deleteMany logs deletes every document in the collection, and updateMany without an update sub-key fails. Pass an explicit filter whenever you mean a subset.
Quote a collection name containing spaces (countDocuments 'my coll'). A collection name starting with -- cannot be expressed — it is read as a flag.
The command line is shell-tokenized but not shell-executed: $, |, >, & and similar shell metacharacters produce an error rather than expanding or redirecting. Wrap values containing them in single quotes.
Unknown flags are rejected rather than ignored; only --db and --query exist.
The scope parameter is not used by MongoDB assets; use --db instead.

Asset config (for put_asset)

field	type	required	notes
`host`	string	yes
`port`	number	yes	No server-side default — pass `27017` explicitly
`username`	string	yes
`password`	string	no	Stored encrypted; never echoed back
`database`	string	no	Default database
`ssh_asset_id`	number	no	SSH asset to tunnel through; 0 detaches