Add a note we do sandbox executions (#282)

note on execution

Co-authored-by: Howon Lee <howon@coderabbit.ai>

Author: howonlee

docs/guides/agent_chat.md

@@ -29,6 +29,8 @@ Upon being invoked, CodeRabbit processes your request and responds with the nece
 
 CodeRabbit Chat can automatically generate scripts in various programming languages (mainly shell scripting) to answer questions about your codebase. If incorrect or non-running or non-germane shell scripts are created, CodeRabbit goes and tries again, and the whole script iteration process will be displayed to you in an Analysis Flow.
 
+All scripts are run in a secure sandboxed execution environment.
+
 ### Web Search
 
 CodeRabbit Chat integrates real-time web search capabilities, enabling it to fetch up-to-date information to support its responses. The CodeRabbit Chat response will indicate this if it decides to search the web to answer your query.

docs/guides/issue-chat.md

@@ -62,6 +62,8 @@ When [agentic thought chain](./agent_chat.md) is enabled, CodeRabbit will use an
 
 ![Agentic Thought Chain Example](../../static/img/guides/agentic-thought-chain.png)
 
+All scripts run in the Agentic Thought Chain are run in a secure sandboxed execution environment.
+
 ### Feature Planning Support
 
 Issue chat is particularly valuable during feature planning phases:

docs/tools/tools.md

@@ -13,6 +13,8 @@ import ProPlanNotice from '@site/src/components/ProPlanNotice.mdx';
 
 CodeRabbit supports various linters and security analysis tools to improve the code review process. The output of these tools is used to enhance the feedback provided by CodeRabbit, making it possible to provide 1-click fixes for common issues.
 
+All tool runs are done in a secure sandboxed execution environment.
+
 ## Enabling/Disabling Tools
 
 You can enable or disable tools by setting `reviews.tools.<tool>.enabled` in your project's `.coderabbit.yaml` file or setting the "Review → Tools → Tool → Enabled" field in CodeRabbit's settings page.