跳到正文
cccskills
免费在 GitHub 公开

add-tavily-tool

Add Tavily web search and URL content extraction to selected NanoClaw agent groups without an API key. Configures MCP, a protocol for connecting agent tools, checks availability, and prepares an upgrade path when the shared allowance runs out.

中文简介正在准备中。当前显示英语简介。

原文查看原文说明

Add Tavily Search and Extract as keyless remote MCP tools for selected NanoClaw agent groups. Use when installing Tavily web search or URL extraction without an API key.

查看安装方法

适用场景

  • Add web search to selected agent groups
  • Extract structured information from URLs
  • Start using Tavily without an API key
  • Prepare an upgrade path for allowance exhaustion

包含的文件(4)

  • SKILL.md7.7 KB
  • REMOVE.md1.3 KB
  • tavily-manifest.test.ts1.3 KB
  • upgrade-instructions.md2.8 KB

SKILL.md(原文)

安装前可以查看交给 AI 代理的指令内容。

Add Tavily Tool

Install the pinned mcp-remote bridge in the agent image and register Tavily's remote MCP server for each selected agent group. The MCP server supplies its tool descriptions and input schemas at runtime.

The registered server exposes:

  • mcp__tavily__tavily_search
  • mcp__tavily__tavily_extract

The registration is provider-agnostic: any provider with MCP support picks it up (Claude, OpenCode, and Codex all do). Groups on the Claude provider already have the built-in WebSearch and WebFetch tools (container/agent-runner/src/providers/claude.ts), so the skill adds the most for groups on other providers, and for Tavily's structured extraction anywhere.

Phase 1: Pre-flight

Check whether the bridge is already in the image manifest, then list the groups:

grep -n '"mcp-remote"' container/cli-tools.json || true
ncl groups list

Ask which agent groups should receive Tavily. If mcp-remote is already present at a pinned version, reuse the existing entry instead of adding a second one.

Phase 2: Install the MCP bridge

Add this object to the top-level array in container/cli-tools.json when an entry named mcp-remote is not already present:

{
  "name": "mcp-remote",
  "version": "0.1.38"
}

Keep the JSON valid and limit the entry to the two fields shown; this package does not require a native build-script opt-in.

Copy the dependency guard into the host test tree:

cp .claude/skills/add-tavily-tool/tavily-manifest.test.ts src/tavily-manifest.test.ts

Build the image and run the guard:

./container/build.sh
pnpm exec vitest run src/tavily-manifest.test.ts

The manifest is the only source-backed integration point. Per-group MCP registration is runtime state stored through ncl, so it has no in-tree line for a registration test to guard.

Phase 3: Register Tavily

config add-mcp-server and groups restart are approval-gated. Run from inside a container they return approval-pending immediately; that is not an error. Wait for the admin's approval and the follow-up system message before moving on to Phase 4.

For each selected <group-id>, register one server named tavily:

ncl groups config add-mcp-server \
  --id <group-id> \
  --name tavily \
  --command mcp-remote \
  --args '["https://mcp.tavily.com/mcp/","--transport","http-only","--enable-proxy","--header","X-Tavily-Access-Mode:keyless","--header","X-Client-Name:nanoclaw","--ignore-tool","tavily_crawl","--ignore-tool","tavily_map","--ignore-tool","tavily_research"]' \
  --env '{}'

The keyless header enables Tavily's IP-based allowance. The client-name header attributes calls to NanoClaw. The tool filters leave only Search and Extract available.

Restart each selected group:

ncl groups restart \
  --id <group-id> \
  --message "Tavily Search and Extract are installed. Run one Tavily search with max_results 1 and report whether it succeeds."

Phase 4: Verify

Confirm the stored configuration contains one tavily server with both headers:

ncl groups config get --id <group-id>

Then check the selected agent's test response. The call must use mcp__tavily__tavily_search. Tavily Crawl, Map, and Research must not appear in the Tavily namespace.

Phase 5: Install the upgrade path

The keyless allowance is shared by every group on the host, so it can run out. Install standing instructions so the agent offers the paid-key upgrade at that moment instead of dead-ending. The upgrade has the user save a key that is then attached to the group through OneCLI agent grants; on an older gateway a saved key would reach every agent. So check the gateway and CLI first. It reads only:

sh .claude/skills/add-onecli/scripts/check-grants-gateway.sh /add-tavily-tool

If it fails, show the operator the message, skip this phase and keep the keyless setup. Also remove an earlier <!-- tavily-upgrade:start --> to <!-- tavily-upgrade:end --> block from each selected group's groups/<group-folder>/instructions.prepend.md, so the agent does not ask anyone to save a key, and restart each group you removed it from (ncl groups restart --id <group-id>): a running agent keeps the old block until it restarts. Install the upgrade path once the gateway is upgraded. Otherwise, for each selected group:

  1. Resolve the OneCLI dashboard URL the user's browser can reach:

    docker inspect onecli --format '{{range .Config.Env}}{{println .}}{{end}}' | grep '^APP_URL='
    

    If the value is a loopback or container-bridge address (127.0.0.1, 172.17.0.1, host.docker.internal), ask the operator which URL they open the OneCLI dashboard at, suggesting http://127.0.0.1:10254 as the default. A public or tailnet APP_URL needs no question.

  2. Gate the deeplink: curl -fs <dashboard-url>/connections/custom must return HTTP 200. If it does not (older OneCLI without the prefill route), replace step 2 of the template with: "Ask an operator to run, on the host: onecli secrets create --name tavily --type generic --host-pattern mcp.tavily.com --header-name Authorization --value-format 'Bearer {value}' --file <key-file>".

  3. Substitute {{ONECLI_DASHBOARD_URL}} and {{AGENT_GROUP_ID}} in upgrade-instructions.md with the resolved URL and the group's id, and write the block into groups/<group-folder>/instructions.prepend.md: replace an existing <!-- tavily-upgrade:start --> to <!-- tavily-upgrade:end --> block in place, append otherwise. Do not write into groups/<group-folder>/CLAUDE.md; it is regenerated at spawn and appended blocks are lost.

  4. Have the operator open the composed deeplink once and confirm the create dialog loads with host mcp.tavily.com prefilled. If they supplied a public URL while APP_URL was a loopback address, suggest setting the public URL in the OneCLI dashboard (Settings, Instance) so future links stay stable.

  5. Restart each selected group: ncl groups restart --id <group-id>.

Keyless limit

If Tavily returns HTTP 429 or monthly_cap_reached_bonus_eligible, the keyless allowance is exhausted. With Phase 5 installed the agent offers the upgrade on its own: the user creates a free API key and stores it through the prefilled dashboard link; the key lands in the OneCLI vault. On OneCLI 1.44+ a new key reaches no agent until it is attached, so the user then attaches it to the group on the host (onecli agents grants attach-secret, a one-liner in the upgrade block), and the gateway injects it into the bridge's requests. The agent then re-registers the server without the X-Tavily-Access-Mode:keyless header and restarts the group. The agent never sees the key.

Troubleshooting

  • command not found: mcp-remote: rebuild the image, then restart the group.
  • Tavily tools are absent: verify the group has a tavily MCP entry, then restart it.
  • Crawl, Map, or Research appears: restore all three --ignore-tool pairs.
  • 429 or monthly_cap_reached_bonus_eligible: the keyless allowance is exhausted; see Keyless limit for the OneCLI upgrade path.
  • The agent reports exhaustion but never offers the upgrade: check that groups/<group-folder>/instructions.prepend.md contains the tavily-upgrade block (Phase 5) and restart the group. A session that already discussed the limit keeps reasoning from that history; /clear starts a clean one.

Removal

See REMOVE.md for the idempotent removal procedure.

References

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护,适用 Google 的隐私权政策和服务条款。

同一仓库的技能

Equip NanoClaw agents to convert attached office documents into Markdown locally using Firecrawl AnyDoc. Agents can read Word, PowerPoint, Excel, and text-based PDF files without uploading them to an external parsing service.

中文简介正在准备中。当前显示英语简介。

  • Reading and summarizing attached office documents
  • Converting documents without uploading them to an external service
  • Extracting text from text-based PDFs
nanocoai/nanoclaw3.1万2026年10月23日 更新

Let a NanoClaw container agent call local models running in Atomic Chat. The skill adds an MCP server, a connection that exposes model listing and text generation as tools, while Claude continues to coordinate the work.

中文简介正在准备中。当前显示英语简介。

  • Delegate part of Claude's work to a local model
  • List models available in Atomic Chat
  • Generate an answer with a specified local model
nanocoai/nanoclaw3.1万2026年10月23日 更新

Add a read-only web dashboard for browsing NanoClaw agents, sessions, and other resources. It builds tables from command-line JSON output and includes message activity charts, recent logs, and a viewer for skills, profiles, and CLAUDE.md files.

中文简介正在准备中。当前显示英语简介。

  • Check agent and session status in a browser
  • Review message counts and daily activity
  • Read recent host logs
nanocoai/nanoclaw3.1万2026年10月23日 更新

add-codex

免费

Add Codex as the agent that plans tasks and coordinates tools in NanoClaw. Choose it for individual groups alongside Claude, with conversation history and session resume using a ChatGPT subscription or OpenAI API key.

中文简介正在准备中。当前显示英语简介。

  • Adding Codex as a NanoClaw agent provider
  • Choosing Claude or Codex for each group
  • Setting up ChatGPT or API key authentication
nanocoai/nanoclaw3.1万2026年10月23日 更新

Install Cursor as an agent backend for NanoClaw groups using its local SDK. Set up authentication and container integration so groups can use Cursor for planning, MCP tools, streaming responses, and session continuation.

中文简介正在准备中。当前显示英语简介。

  • Install Cursor as a NanoClaw agent backend
  • Configure Cursor account sign-in or API key authentication
  • Switch a specific NanoClaw group to Cursor
nanocoai/nanoclaw3.1万2026年10月23日 更新

Install a local NanoClaw dashboard to inspect agents, sessions, channels, messages, and users. It displays token usage, context usage, activity, and live logs using periodic data snapshots and a log forwarding module.

中文简介正在准备中。当前显示英语简介。

  • Checking agent and session status in one place
  • Reviewing token and context usage
  • Following live logs with level filters
nanocoai/nanoclaw3.1万2026年10月23日 更新

查看 nanocoai 的全部技能

报告此技能的问题