一键导入
cli-default
CLI tool (agentic-db) for the default API — typed CRUD, unified search, and context/config commands for the 91 generated tables.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
CLI tool (agentic-db) for the default API — typed CRUD, unified search, and context/config commands for the 91 generated tables.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
CLI tool (agentic-db) for the default API — provides CRUD commands for 95 tables and 3 custom operations
ORM client for the default API — provides typed CRUD operations for 95 tables and 3 custom operations
ORM client for the default API — typed CRUD, vector + hybrid search, and relation walking for the 91 generated tables.
Install and deploy the agentic-db PostgreSQL module using pgpm. Use when asked to 'install agentic-db', 'deploy the database', 'set up agentic-db', 'create a workspace', or when bootstrapping a new project that depends on agentic-db.
| name | cli-default |
| description | CLI tool (agentic-db) for the default API — typed CRUD, unified search, and context/config commands for the 91 generated tables. |
The agentic-db CLI wraps the generated ORM with a fluent per-table command
surface plus a small set of hand-authored top-level commands (context,
config, auth, search, ask, init, gen, request-upload-url,
confirm-upload, provision-bucket).
Per-entity commands are auto-generated from the GraphQL schema — there is one
subcommand per table in the generated ORM. See
sdk/cli/generated/cli/commands/ for
the authoritative list; common ones include contact, company, note,
task, memory, conversation, message, agent, skill, rule,
project, goal, habit, and their M:N junctions (contact-company,
contact-note, task-project, goal-habit, …).
# Context management
agentic-db context create <name> --endpoint <url>
agentic-db context use <name>
agentic-db context list
# Authentication
agentic-db auth set-token <token>
# Config variables (provider, tty, format, …)
agentic-db config set <key> <value>
agentic-db config show
# CRUD for any generated table (e.g. contact)
agentic-db contact list --select id,firstName,lastName --json --tty false
agentic-db contact get --id <uuid> --select id,firstName --tty false
agentic-db contact create --firstName Alice --lastName Smith --select id --tty false
agentic-db contact update --id <uuid> --headline "Staff Engineer" --tty false
agentic-db contact delete --id <uuid> --tty false
# Unified search (vector + full-text) over one or more tables
agentic-db search "postgres distributed systems" \
--tables contacts,memories,notes \
--json --tty false
# RAG question-answering across tables
agentic-db ask "Who did I meet about the Q2 launch?" --tty false
agentic-db context create local --endpoint http://localhost:5000/graphql
agentic-db context use local
agentic-db auth set-token $AGENTIC_DB_TOKEN
agentic-db contact list --select id,firstName,lastName --json --tty false
All commands accept --tty false (or --no-tty) to skip prompts and emit
JSON-only output suitable for piping:
agentic-db contact create \
--firstName Frida --lastName Finch \
--headline "CLI smoke test" \
--select id,firstName --json --tty false
Filters use the same where.<field>.<op> flag syntax the ORM exposes:
agentic-db contact list \
--where.email.likeInsensitive '%@example.com' \
--select id,firstName,email \
--json --tty false
The end-to-end test suite in
packages/cli-e2e-tests/__tests__/cli-e2e.test.ts
pins the behavior of every example above — config show, context list,
search --tables tasks, contact list, note list, and a contact create
round-trip. If you add a new hand-authored top-level command, add a test in
that file so this skill and the CLI stay aligned.