Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add DemonDamon/AgenticX --skill connector-assistantgit clone --depth 1 https://github.com/DemonDamon/AgenticXWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/demondamon/agenticx/connector-assistant)<a href="https://agentmods.dev/skills/demondamon/agenticx/connector-assistant"><img src="https://agentmods.dev/badge/skills/demondamon/agenticx/connector-assistant/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/demondamon/agenticx/connector-assistant"><img src="https://agentmods.dev/badge/skills/demondamon/agenticx/connector-assistant.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00110 | $0.03789 |
| Opus 5.5 | $0.00044 | $0.01516 |
| Sonnet 5.5 | $0.00022 | $0.00758 |
| Haiku 4.5 | $0.00011 | $0.00379 |
Grade A, and why
connector-assistant scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 159 lines — stays where its author put it; the contents beside it link to each section on GitHub.
连接器助手
0. 概念:连接器 ≠ 分身
- 连接器(Connector)=一个「模板实例」:把外部系统接进 Near,在「设置 → 连接器 → 我的连接」、市场卡片、对话框连接器选择器里只出现一次。三种形态:
mcp:远程 MCP Server(存于~/.agenticx/mcp.json);rest:HTTP API,登记到本机连接器网关(connector-runtime,~/.agenticx/connector-runtime/connectors/<id>.json),对话里通过网关 MCP 工具search_actions → get_action_guide → execute_action调用,实例 id 形如rest:<id>;database:SQLite / MySQL / PostgreSQL,内置只读 MCP(list_tables/describe_table/query),存于 mcp.json。
- 分身(Avatar)=另一个智能体人格,与连接器无关。
- 通用 MCP Server ≠ 连接器:手动添加 / 导入 / 市场安装的 MCP 只在「插件市场 → MCP」里展示与管理,不进「我的连接」。用户只是想加一个 MCP Server(不要求作为连接器)时,不走本助手,引导到「插件市场 → MCP → + 新建 MCP」(远程 URL / 导入 JSON);「新建连接器」菜单里没有「新建自定义 MCP」。只有用户明确要把它作为连接器接入时,才用
connector_manage create建(会打连接器标,出现在「我的连接」)。 - 用户说「创建一个连接器 叫 abc」→
abc是连接器名。禁止create_avatar、禁止追问「abc 是不是分身」、禁止说「无法创建原生连接器」然后结束。
1. 工具
只用 connector_manage(不要手写 / file_edit mcp.json,不要 mcp_import):
| action | 用途 |
|---|---|
list_templates |
模板目录(create_via: mcp_url 可在对话里建;ui_native 需在 UI 握手;unavailable 未接线) |
list_instances |
已有实例:instances(mcp / database)+ rest_instances(网关 REST),只返回 has_credential,不含密钥 |
create |
新建 / 更新 MCP 实例(mcp_oauth:true = 标准 MCP OAuth,verify 时浏览器授权) |
create_rest |
登记 REST 连接器(OpenAPI/Swagger URL 或文件优先;否则简短接口清单) |
create_database |
新建数据库连接器(默认只读 + 行数上限) |
request_credential |
弹出掩码卡片按类型收凭证(多字段),你拿不到明文;OAuth 授权码模式会在填完后自动打开系统浏览器授权 |
authorize_oauth |
重新走 OAuth 授权码(PKCE)浏览器授权(REST) |
verify |
探活:MCP 握手 / REST 执行一个无参 read 动作 / 数据库连接 + 列表 |
delete |
仅在用户明确要求删除时使用 |
2. 流程(严格按序)
第 1 步:先盘点,避免重复建设
- 并行调用
connector_manage {action:"list_templates"}与{action:"list_instances"}。 - 按名称 / 关键词匹配:
- 已有同名或同系统实例 → 告诉用户已存在,用
request_clarification问「直接使用 / 更新连接」(单题即可)。选「更新」时create带overwrite:true。 - 命中模板且
create_via=mcp_url→ 走模板实例:template_id用该模板 id。模板带mcp_url(官方端点,已核实)时不要再问地址,url可省略(自动用mcp_url);否则只问上游地址(第 2 步只保留 Q3)。认证方式按模板auth:mcp_oauth(官方远程 MCP,OAuth 2.1 + 动态客户端注册,如腾讯文档 / Notion / Linear / Cloudflare / 天眼查 / 企查查)→ 只需名称,绝不索要 Token / Client ID / Secret;create 自动mcp_oauth:true,随后verify打开系统浏览器授权,令牌只存本机~/.agenticx/connectors/oauth/(0600)。- 模板带
auth_query(如高德key、百度地图ak、腾讯地图key、快递100key)→ create 自动auth_style:"query"+query_param,再request_credential用掩码卡片收 Key。 - 模板带
auth_header(如盈米x-api-key)→ create 自动auth_style:"header"+header_name,同样用掩码卡片收 Key。 api_key/custom_credential→bearer(模板 create 未传auth_style时自动 bearer);none→none。- 模板带
credential_label/credential_help_url(如恒生聚源:Access Token+ 飞书「如何获取凭证」文档)→ 收凭证前在回复里用一句话附上该链接(Markdown 链接,如「如何获取凭证」),request_credential的credential_label用模板给的名称(不传也会自动取模板值)。
- 命中
ui_native模板(腾讯会议 / TAPD / GitHub / 飞书 / 企业微信 / Agent Mail)→ 不在对话里建,引导用户到「市场 / 设置 → 连接器」点「连接」完成授权;结束。 - 命中
unavailable模板(OAuth 类)→ 说明暂未接线,可改用该服务的 MCP Server URL 走自定义。 - 都不匹配 → 第 2 步。
- 已有同名或同系统实例 → 告诉用户已存在,用
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 7d ago First seen · 159 lines · 110 tokens per session scan A 5a0b1ee10749
connector-assistant is a skill published in the GitHub repository DemonDamon/AgenticX (443 stars, last pushed today), licensed Apache-2.0. It adds 110 tokens to every session and 3,789 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-09.
Other skills, from other repositories
continuum-tools-mcp
Connect MCP servers (Stdio/SSE/StreamableHTTP) to a Continuum agent, configure tool filtering, set up tool-context capture/injection (e.g. sessionid), and read run artifacts (UI widgets, structured tool data). Invoke when the user asks "connect MCP", "filesystem tool", "remote API tool", "auto-capture sessionid"…
citedy-content-ingestion
Turn any URL into structured content — YouTube videos (via Gemini Video API), web articles, PDFs, and audio files. Extract transcripts, summaries, and metadata for use in any LLM pipeline. Powered by Citedy.
V3 MCP Optimization
MCP server optimization and transport layer enhancement for claude-flow v3. Implements connection pooling, load balancing, tool registry optimization, and performance monitoring for sub-100ms response times.
weather
Get current weather and forecasts.
github
Manage GitHub issues and pull requests, inspect CI runs, and query repository data.
ai-persistence/server
Server chat state with withPersistence from @tanstack/ai-persistence. Authoritative transcript, run lifecycle, durable interrupts/approvals, chatParamsFromRequest, reconstructChat, snapshotStreaming. Use when the server owns history, multi-device, or durable tool approvals. NOT client localStorage (see…