update

Author: Palashio

src/langsmith/skills.mdx

@@ -3,7 +3,7 @@ title: LangSmith skills
 description: Use Agent Skills to work with LangSmith traces, datasets, and evaluators from your coding agent.
 ---
 
-Agent Skills are reusable, on‑demand capabilities that bundle instructions plus optional helper scripts. This page summarizes the LangSmith‑oriented skills you can add to a compatible coding agent to query traces, generate datasets, and define evaluators.
+Agent Skills are reusable, on‑demand capabilities that bundle instructions plus optional helper scripts. This page summarizes the LangSmith‑oriented skills you can add to a compatible coding agent to query traces, generate datasets, and define evaluators. To work with the same LangSmith data directly from the terminal, use the [LangSmith CLI](/langsmith/langsmith-cli).
 
 <Note>
 These skills follow the Agent Skills specification and are provided in the langchain‑skills package. You can copy the `SKILL.md` and any referenced `scripts/` into your agent’s skills directory. The installers below only install the LangSmith skills (trace, dataset, evaluator).
@@ -16,7 +16,8 @@ Only the LangSmith skills (trace, dataset, evaluator):
 <Tabs>
   <Tab title="Claude Code">
 
-  ### Local installation
+  ### Installation
+  Install locally (add `--global` to install globally):
   <CodeGroup>
   ```bash [macOS/Linux]
   curl -fsSL https://github.com/langchain-ai/langchain-skills/archive/main.tar.gz \
@@ -32,26 +33,13 @@ Only the LangSmith skills (trace, dataset, evaluator):
   ```
   </CodeGroup>
 
-  ### Global installation
-  <CodeGroup>
-  ```bash [macOS/Linux]
-  curl -fsSL https://github.com/langchain-ai/langchain-skills/archive/main.tar.gz \
-    | tar -xz -C /tmp \
-    && /tmp/langchain-skills-main/install.sh --global --langsmith -y \
-    && rm -rf /tmp/langchain-skills-main
-  ```
-  ```bash [Windows (WSL/Git Bash)]
-  curl -fsSL https://github.com/langchain-ai/langchain-skills/archive/main.tar.gz \
-    | tar -xz -C /tmp \
-    && /tmp/langchain-skills-main/install.sh --global --langsmith -y \
-    && rm -rf /tmp/langchain-skills-main
-  ```
-  </CodeGroup>
+  For a global install, add `--global` to the \`install.sh\` invocation (for example \`install.sh --global --langsmith -y\`).
 
   </Tab>
   <Tab title="DeepAgents CLI">
 
-  ### Local installation
+  ### Installation
+  Install locally (add `--global` to install globally):
   <CodeGroup>
   ```bash [macOS/Linux]
   curl -fsSL https://github.com/langchain-ai/langchain-skills/archive/main.tar.gz \
@@ -67,21 +55,7 @@ Only the LangSmith skills (trace, dataset, evaluator):
   ```
   </CodeGroup>
 
-  ### Global installation
-  <CodeGroup>
-  ```bash [macOS/Linux]
-  curl -fsSL https://github.com/langchain-ai/langchain-skills/archive/main.tar.gz \
-    | tar -xz -C /tmp \
-    && /tmp/langchain-skills-main/install.sh --deepagents --global --langsmith -y \
-    && rm -rf /tmp/langchain-skills-main
-  ```
-  ```bash [Windows (WSL/Git Bash)]
-  curl -fsSL https://github.com/langchain-ai/langchain-skills/archive/main.tar.gz \
-    | tar -xz -C /tmp \
-    && /tmp/langchain-skills-main/install.sh --deepagents --global --langsmith -y \
-    && rm -rf /tmp/langchain-skills-main
-  ```
-  </CodeGroup>
+  For a global install, add `--global` to the \`install.sh\` invocation (for example \`install.sh --deepagents --global --langsmith -y\`).
 
   </Tab>
 </Tabs>
@@ -92,13 +66,14 @@ To update, re‑run the command. If target skill folders already exist, remove t
 
 ## Configure environment
 
-After installing the skills, set environment variables used by all LangSmith skills and helper scripts:
+After installing the skills, set environment variables used by all LangSmith skills, helper scripts, and the [LangSmith CLI](/langsmith/langsmith-cli):
 
 ```bash
 export LANGSMITH_API_KEY=<your-key>
-# Optional
+# Optional defaults
 export LANGSMITH_PROJECT=<default-project>
-export LANGSMITH_WORKSPACE_ID=<workspace-id>
+# Advanced: multi-workspace or certain self-hosted setups only
+# export LANGSMITH_WORKSPACE_ID=<workspace-id>
 ```
 
 ## What these skills cover
@@ -107,7 +82,7 @@ export LANGSMITH_WORKSPACE_ID=<workspace-id>
 - Datasets: Turn traces into evaluation datasets (final_response, single_step, trajectory, RAG) and optionally upload to LangSmith.
 - Evaluators: Define code or LLM‑as‑judge evaluators and attach them to datasets (offline) or projects (online).
 
-Each skill directory ships with a `SKILL.md` plus optional `scripts/` helpers you can run or adapt.
+Each skill directory ships with a `SKILL.md` plus optional `scripts/` helpers you can run or adapt. These skills are designed to plug into compatible coding agents (such as Claude Code or DeepAgents CLI), though you can also reuse the helper scripts directly if you prefer not to wire up a full agent.
 
 ## Manual install