From 95e45b370cb596bacb59f2c9e21a2bda496a0f64 Mon Sep 17 00:00:00 2001 From: Thierry Damiba <177760997+thierrypdamiba@users.noreply.github.com> Date: Wed, 29 Jul 2026 13:01:25 -0700 Subject: [PATCH 1/4] fix(integrations): correct Nimble page capabilities and token-usage claims MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Follow-up to #1097. - Add the web search agent capability (Nimble documents six tools; the page listed five) and rename "Templates" to "Extract templates" to match Nimble's naming. - Replace the "Caching and cost" section. It described a cache and memory layer that "avoids redundant searches and the token cost they add" — none of which appears anywhere in Nimble's docs, and which contradicts Nimble's own MCP page warning that MCP "can consume tokens quickly." Replaced with accurate token-usage guidance and a pointer to the CLI skill Nimble recommends for token efficiency. - Show the real namespaced tool name (nimble.nimble_search) rather than a . placeholder, since Nimble prefixes its tools with nimble_ and the doubled prefix is surprising. - Changelog: "MCP Server" -> "MCP server" per STYLEGUIDE.md. Co-Authored-By: Claude Opus 5 (1M context) --- app/en/references/changelog/page.mdx | 2 +- .../resources/integrations/search/nimble/page.mdx | 14 ++++++++++---- 2 files changed, 11 insertions(+), 5 deletions(-) diff --git a/app/en/references/changelog/page.mdx b/app/en/references/changelog/page.mdx index ac219ae7c..bb3003eac 100644 --- a/app/en/references/changelog/page.mdx +++ b/app/en/references/changelog/page.mdx @@ -13,7 +13,7 @@ _Here's what's new at Arcade.dev!_ **Misc** -- `[feature - 🚀]` Add Nimble as a Partner MCP Server for real-time web data in the search integrations catalog. +- `[feature - 🚀]` Add Nimble as a Partner MCP server for real-time web data in the search integrations catalog. ## 2026-07-24 diff --git a/app/en/resources/integrations/search/nimble/page.mdx b/app/en/resources/integrations/search/nimble/page.mdx index e7da5587c..c1693a816 100644 --- a/app/en/resources/integrations/search/nimble/page.mdx +++ b/app/en/resources/integrations/search/nimble/page.mdx @@ -40,15 +40,21 @@ Once you register Nimble in your Arcade project, Arcade discovers its tools auto | Extract | Pull structured content from a specific URL, with sync and async extraction. | | Crawl | Crawl multiple pages with path filtering, subdomain control, and progress tracking. | | Map | Discover the URLs on a site through link-following and sitemaps. | -| Templates | Run pre-built extraction templates for common websites. | +| Extract templates | Browse, run, and generate pre-built extraction templates for popular websites. | +| Web search agent | Create, run, and poll research agents, and read their cited results. | -Arcade namespaces each tool under the server ID you choose when you add the server, so tool names appear as `.` in the Playground and gateway tool picker. +Nimble prefixes its own tools with `nimble_`, and Arcade namespaces each tool under the server ID you choose when you add the server. With the server ID `nimble`, tool names appear as `nimble.nimble_search` in the Playground, the gateway tool picker, and SDK calls. Copy the exact name from the tool picker before calling it. Compose these tools with Google Docs, Slack, Salesforce, GitHub, or any of Arcade's native servers in a single MCP Gateway, so an agent can research, draft, and act in one flow, with authorization and audit from Arcade's runtime. -## Caching and cost +## Token usage -Nimble's tools sit behind a cache and a memory layer. The cache stores sources, search history, and responses; the memory layer extracts key learnings after each run and feeds them into the next search plan. Together they retrieve only what an agent needs, which avoids redundant searches and the token cost they add, and refines results for a given domain over time. This behavior runs server-side at Nimble. When you call these tools through an Arcade gateway, Arcade adds per-user authorization and audit on top. +Web retrieval returns large payloads, and MCP routes every result through the model's context window, so these tools can consume tokens quickly. Two things help: + +- **Expose only the tools you need.** An MCP Gateway sends just the tools you select, so an agent that only searches never loads the crawl and template tool definitions. +- **Prefer the narrowest tool.** Reach for Search or Extract before Crawl or the web search agent, which return content across many pages. + +Nimble also documents a [CLI skill](https://docs.nimbleway.com/integrations/agent-skills/web-tools-skills/nimble-web-expert) that it recommends over MCP for typical web retrieval on token-efficiency grounds. Use the MCP server when you want Nimble governed alongside your other tools, with Arcade handling per-user authorization and audit at runtime. ## Add Nimble to your Arcade project From b06f6250fde4c5510fe7481c3e4b49c63ec4dd70 Mon Sep 17 00:00:00 2001 From: Thierry Damiba <177760997+thierrypdamiba@users.noreply.github.com> Date: Wed, 29 Jul 2026 15:41:59 -0700 Subject: [PATCH 2/4] fix(integrations): restore Nimble cost section with sourced claims MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit I removed this section on the grounds that Nimble's docs didn't support it. That was wrong — I only checked docs.nimbleway.com. The claims are published on nimbleway.com/mcp and in Nimble's MCP blog post: "Compound accuracy over time through a Proprietary Index and Memory that gets smarter with every query" "Cut token costs by retrieving exactly what's needed — no redundant searches, no parsing raw pages with an LLM" Restored, with two changes from the original wording. The claims are now attributed to Nimble rather than asserted as fact, since they're product claims we can't verify from outside. And the mechanism detail the original added beyond the source (that the cache stores "sources, search history, and responses" and the memory layer "extracts key learnings after each run") is dropped — Nimble publishes "Proprietary Index and Memory", not that breakdown. Kept the practical MCP token caveat alongside it. Both are true and in tension: the marketing claims retrieval efficiency, while Nimble's docs warn MCP "can consume tokens quickly" and point high-volume users at their CLI skill. Readers should see both. Also verified tool names against Nimble's docs: nimble_search is correct, so nimble.nimble_search stands. Co-Authored-By: Claude Opus 5 (1M context) --- app/en/resources/integrations/search/nimble/page.mdx | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/app/en/resources/integrations/search/nimble/page.mdx b/app/en/resources/integrations/search/nimble/page.mdx index c1693a816..ce4d22dbd 100644 --- a/app/en/resources/integrations/search/nimble/page.mdx +++ b/app/en/resources/integrations/search/nimble/page.mdx @@ -47,14 +47,16 @@ Nimble prefixes its own tools with `nimble_`, and Arcade namespaces each tool un Compose these tools with Google Docs, Slack, Salesforce, GitHub, or any of Arcade's native servers in a single MCP Gateway, so an agent can research, draft, and act in one flow, with authorization and audit from Arcade's runtime. -## Token usage +## Cost and token usage -Web retrieval returns large payloads, and MCP routes every result through the model's context window, so these tools can consume tokens quickly. Two things help: +Nimble designs its [MCP server](https://www.nimbleway.com/mcp) around retrieval efficiency. It describes a proprietary index and memory that "gets smarter with every query" to compound accuracy over time, and cutting token costs "by retrieving exactly what's needed," without redundant searches or parsing raw pages through an LLM. Its tools return schema-consistent JSON, so your model spends tokens reasoning over results rather than parsing raw HTML. This behavior runs server-side at Nimble. + +Web retrieval still returns large payloads, and MCP routes every result through the model's context window. Two things keep that in check: - **Expose only the tools you need.** An MCP Gateway sends just the tools you select, so an agent that only searches never loads the crawl and template tool definitions. -- **Prefer the narrowest tool.** Reach for Search or Extract before Crawl or the web search agent, which return content across many pages. +- **Prefer the narrowest tool.** Reach for search or extract before crawl or the web search agent, which return content across many pages. -Nimble also documents a [CLI skill](https://docs.nimbleway.com/integrations/agent-skills/web-tools-skills/nimble-web-expert) that it recommends over MCP for typical web retrieval on token-efficiency grounds. Use the MCP server when you want Nimble governed alongside your other tools, with Arcade handling per-user authorization and audit at runtime. +For high-volume retrieval, Nimble also documents a [CLI skill](https://docs.nimbleway.com/integrations/agent-skills/web-tools-skills/nimble-web-expert) as a more token-efficient path than MCP. Use the MCP server when you want Nimble governed alongside your other tools, with Arcade adding per-user authorization and audit at runtime. ## Add Nimble to your Arcade project From 647458840616b4f7564789cd2c55cc4a0812a447 Mon Sep 17 00:00:00 2001 From: Thierry Damiba <177760997+thierrypdamiba@users.noreply.github.com> Date: Wed, 29 Jul 2026 16:16:28 -0700 Subject: [PATCH 3/4] fix(integrations): correct tool list against live Nimble MCP server MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Verified against mcp.nimbleway.com/mcp (Nimble 3.4.3, protocol 2025-06-18) via a tools/list handshake rather than published docs. The server exposes 27 tools, not 6. Nimble's docs describe six capability families; each ships multiple tools for sync, async, and status/listing calls. This matters in Arcade specifically, because the gateway tool picker lists individual tools — someone reading "6 capabilities" then seeing 27 checkboxes has no way to map between them. Table now carries a per-capability tool count and the total. Added a callout on the async pattern. Many tools return a task ID and require polling a companion tool (nimble_task_results, nimble_agents_run_status). A gateway built with only the starting tools selected will kick off jobs the agent cannot read back. Worth calling out since selecting a subset of tools is the normal gateway workflow. Resolved the open question from the previous commit: the live server has no Google Maps tools and no nimble_deep_web_search, so the tool names on nimbleway.com/mcp are stale. Docs were right, marketing page was not. nimble_search confirmed present, so nimble.nimble_search stands. Co-Authored-By: Claude Opus 5 (1M context) --- .../integrations/search/nimble/page.mdx | 29 ++++++++++++------- 1 file changed, 19 insertions(+), 10 deletions(-) diff --git a/app/en/resources/integrations/search/nimble/page.mdx b/app/en/resources/integrations/search/nimble/page.mdx index ce4d22dbd..0c5476000 100644 --- a/app/en/resources/integrations/search/nimble/page.mdx +++ b/app/en/resources/integrations/search/nimble/page.mdx @@ -32,19 +32,28 @@ Set these under **Advanced settings → Custom headers** when you add the server ## What you can do -Once you register Nimble in your Arcade project, Arcade discovers its tools automatically. They cover these capabilities: - -| Capability | What it does | -| --- | --- | -| Search | Real-time web search across multiple engines, with configurable depth and focus. | -| Extract | Pull structured content from a specific URL, with sync and async extraction. | -| Crawl | Crawl multiple pages with path filtering, subdomain control, and progress tracking. | -| Map | Discover the URLs on a site through link-following and sitemaps. | -| Extract templates | Browse, run, and generate pre-built extraction templates for popular websites. | -| Web search agent | Create, run, and poll research agents, and read their cited results. | +Once you register Nimble in your Arcade project, Arcade discovers its tools automatically. Nimble exposes 27 tools, because most capabilities ship more than one tool: a synchronous call, an asynchronous variant, and status or listing calls. They group into these capabilities: + +| Capability | Tools | What it does | +| --- | --- | --- | +| Search | 1 | Real-time web search with configurable content richness. | +| Extract | 2 | Pull structured content from a specific URL, synchronously or asynchronously. | +| Crawl | 4 | Crawl multiple pages on a site, then check progress, list jobs, or cancel a run. | +| Map | 1 | Discover the URLs on a site by crawling its pages and sitemap. | +| Extract templates | 7 | Browse the template catalog, run templates against a URL, and generate or refine custom ones. | +| Web search agents | 11 | Create, update, and run research agents, then poll runs and read their results. | +| Task results | 1 | Fetch the status and results of any async task. | Nimble prefixes its own tools with `nimble_`, and Arcade namespaces each tool under the server ID you choose when you add the server. With the server ID `nimble`, tool names appear as `nimble.nimble_search` in the Playground, the gateway tool picker, and SDK calls. Copy the exact name from the tool picker before calling it. + + Many Nimble tools are asynchronous: they return a task ID immediately, and + your agent polls a companion tool such as `nimble_task_results` or + `nimble_agents_run_status` for the outcome. When you build a gateway, select + the polling tools alongside the tools that start the work, or your agent will + kick off jobs it cannot read back. + + Compose these tools with Google Docs, Slack, Salesforce, GitHub, or any of Arcade's native servers in a single MCP Gateway, so an agent can research, draft, and act in one flow, with authorization and audit from Arcade's runtime. ## Cost and token usage From b1d6e47ef5ad844c813123a501ab8905571abc87 Mon Sep 17 00:00:00 2001 From: Thierry Damiba <177760997+thierrypdamiba@users.noreply.github.com> Date: Wed, 29 Jul 2026 16:35:46 -0700 Subject: [PATCH 4/4] fix(integrations): stop asserting an exact Nimble tool count The live server returned 27 tools for our account, but that number does not belong in the docs as a fact. tools/list is auth-gated: requests without a valid key get 401, so I could not compare what a different plan or account sees. Our access is a partner-provided trial, which is exactly the kind of account likely to differ from a customer's. Arcade's own add-remote-servers guide documents this as expected behavior: Arcade pre-loads the tools available to the user who configures the server, and a tool unavailable to an end-user will not appear via the gateway. So the number is per-account by design. Now gives the reader the magnitude that matters (expect roughly two dozen tools, not seven), says the set depends on plan and account, and points at the guide's admin-user guidance. Capability families stay, since those are stable across Nimble's docs and the live server. Also added the read/write split. Most tools are read-only; the agent create/update/run/delete and template-generation tools write to your Nimble account, and the server marks agent deletion destructive. Worth knowing before you drop the whole toolset into a research gateway. Co-Authored-By: Claude Opus 5 (1M context) --- .../integrations/search/nimble/page.mdx | 26 +++++++++++-------- 1 file changed, 15 insertions(+), 11 deletions(-) diff --git a/app/en/resources/integrations/search/nimble/page.mdx b/app/en/resources/integrations/search/nimble/page.mdx index 0c5476000..f7aab9a61 100644 --- a/app/en/resources/integrations/search/nimble/page.mdx +++ b/app/en/resources/integrations/search/nimble/page.mdx @@ -32,17 +32,19 @@ Set these under **Advanced settings → Custom headers** when you add the server ## What you can do -Once you register Nimble in your Arcade project, Arcade discovers its tools automatically. Nimble exposes 27 tools, because most capabilities ship more than one tool: a synchronous call, an asynchronous variant, and status or listing calls. They group into these capabilities: - -| Capability | Tools | What it does | -| --- | --- | --- | -| Search | 1 | Real-time web search with configurable content richness. | -| Extract | 2 | Pull structured content from a specific URL, synchronously or asynchronously. | -| Crawl | 4 | Crawl multiple pages on a site, then check progress, list jobs, or cancel a run. | -| Map | 1 | Discover the URLs on a site by crawling its pages and sitemap. | -| Extract templates | 7 | Browse the template catalog, run templates against a URL, and generate or refine custom ones. | -| Web search agents | 11 | Create, update, and run research agents, then poll runs and read their results. | -| Task results | 1 | Fetch the status and results of any async task. | +Once you register Nimble in your Arcade project, Arcade discovers its tools automatically. Nimble groups them into these capabilities: + +| Capability | What it does | +| --- | --- | +| Search | Real-time web search with configurable content richness. | +| Extract | Pull structured content from a specific URL, synchronously or asynchronously. | +| Crawl | Crawl multiple pages on a site, then check progress, list jobs, or cancel a run. | +| Map | Discover the URLs on a site by crawling its pages and sitemap. | +| Extract templates | Browse the template catalog, run templates against a URL, and generate or refine custom ones. | +| Web search agents | Create, update, and run research agents, then poll runs and read their results. | +| Task results | Fetch the status and results of any async task. | + +Most capabilities ship more than one tool: a synchronous call, an asynchronous variant, and status or listing calls. Expect roughly two dozen individual tools in the tool picker rather than seven. The exact set depends on your Nimble plan and on the account you connect with, so [connect as an admin user](/guides/mcp-gateways/add-remote-servers#save-and-confirm-the-connection) to discover the broadest list, and check the picker for what your project actually has. Nimble prefixes its own tools with `nimble_`, and Arcade namespaces each tool under the server ID you choose when you add the server. With the server ID `nimble`, tool names appear as `nimble.nimble_search` in the Playground, the gateway tool picker, and SDK calls. Copy the exact name from the tool picker before calling it. @@ -54,6 +56,8 @@ Nimble prefixes its own tools with `nimble_`, and Arcade namespaces each tool un kick off jobs it cannot read back. +Most Nimble tools only read. The exceptions write to your Nimble account: the tools that create, update, run, or delete agents, and the ones that generate extract templates. If you are building a research gateway, leave the write tools out of it. + Compose these tools with Google Docs, Slack, Salesforce, GitHub, or any of Arcade's native servers in a single MCP Gateway, so an agent can research, draft, and act in one flow, with authorization and audit from Arcade's runtime. ## Cost and token usage