docs: add Vaultak runtime security integration for CrewAI agents

[samueloladji-beep]

Summary

Adds a new observability page for Vaultak — a runtime security platform that wraps CrewAI agents with real-time protection.

• docs/en/observability/vaultak.mdx — step-by-step integration guide covering:
  • What Vaultak does and why it matters for agentic workloads
  • 4-step setup with a complete multi-agent crew example
  • Configuration reference table (all 5 constructor parameters)
  • Monitored-event table (what Vaultak does at each CrewAI/LangChain lifecycle hook)
  • Stricter-threshold example for sensitive workloads (databases, finance)
• docs/en/observability/overview.mdx — adds a new "Runtime Security" <CardGroup> section with a Vaultak card
• docs/docs.json — registers en/observability/vaultak in the Observability nav group across all language/version entries

Install

pip install langchain-vaultak

Quick preview

from crewai import Agent, Task, Crew, Process
from langchain_openai import ChatOpenAI
from langchain_vaultak import VaultakCallbackHandler

vaultak = VaultakCallbackHandler(api_key="vtk_...", agent_name="my-crew")
llm = ChatOpenAI(model="gpt-4o", callbacks=[vaultak])

agent = Agent(role="Analyst", goal="...", backstory="...", llm=llm)
crew = Crew(agents=[agent], tasks=[task], process=Process.sequential)
crew.kickoff()

Test plan

• Page renders correctly at /en/observability/vaultak
• Overview card links correctly to the new page
• Nav sidebar shows Vaultak under Observability
• Code examples are syntactically correct

🤖 Generated with Claude Code

Summary by CodeRabbit

• Documentation
  • Added a new Vaultak Integration page describing runtime security and monitoring for agents.
  • Added a Runtime Security subsection to the observability overview with a Vaultak card and link.
  • Docs cover features (risk scoring, policy checks, PII masking, blocking/rollback, alerts, session lifecycle), configurable options, dashboard overview, integration steps, and links to resources.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 3fc832d6-fdf8-48e8-9309-2a634fa4605f

📥 Commits

Reviewing files that changed from the base of the PR and between 5e1ea46 and ccb835b.

📒 Files selected for processing (1)

• docs/en/observability/vaultak.mdx

✅ Files skipped from review due to trivial changes (1)

• docs/en/observability/vaultak.mdx

---

📝 Walkthrough

Walkthrough

Adds Vaultak runtime security docs: a Runtime Security card on the observability overview linking to a new Vaultak Integration page that documents installation, handler usage with a Python example, configuration options, monitored events, and reference links.

Changes

Vaultak Runtime Security Integration Documentation

Layer / File(s)	Summary
Overview navigation to Vaultak docs/en/observability/overview.mdx	Adds a Runtime Security subsection with a Vaultak card that links to the new integration documentation page.
Vaultak documentation introduction and context docs/en/observability/vaultak.mdx	New integration page with frontmatter metadata and introduction explaining Vaultak's real-time interception, risk scoring, and runtime security capabilities for CrewAI agents.
Getting started and configuration docs/en/observability/vaultak.mdx	Step-by-step setup (sign-up/API key, pip install langchain-vaultak, handler attachment to an LLM) with a complete Python code example. Includes configuration section documenting handler parameters and threshold customization for sensitive workloads.
Monitoring capabilities and references docs/en/observability/vaultak.mdx	Table mapping CrewAI events to Vaultak monitoring actions (session lifecycle, risk scoring/blocking, policy enforcement, PII masking, alerts, rollback), plus links to documentation, PyPI, GitHub, and the Vaultak dashboard.

"A security friend named Vaultak arrives,
Runtime risks caught in its quick-moving grips!
Risk scores and masks, policies kept tight,
Agents roam safer in the dim of night. 🐇✨"

🎯 1 (Trivial) | ⏱️ ~3 minutes

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The pull request title accurately describes the main change: adding documentation for Vaultak runtime security integration for CrewAI agents, which aligns with the file changes and PR objectives.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

🧹 Nitpick comments (2)

docs/en/observability/vaultak.mdx (2)

112-121: 💤 Low value

Consider demonstrating the verbose parameter in the code example.

The configuration table documents five parameters including verbose, but the code example (lines 56-61) only demonstrates four. While not incorrect, adding verbose=True to the example would make the documentation more comprehensive.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@docs/en/observability/vaultak.mdx` around lines 112 - 121, Update the example
configuration block to include the verbose parameter so readers see how to
enable verbose logging; specifically add verbose=True alongside the existing
parameters (e.g., where the example constructs the Vaultak client/config with
api_key and agent_name) so the code snippet demonstrates all five documented
options including verbose.

---

1-6: ⚡ Quick win

Icon mismatch between overview card and page frontmatter.

The overview page uses icon="shield-halved" for the Vaultak card (line 66 in overview.mdx), but this page's frontmatter specifies icon="shield-check". For consistency, both should use the same icon.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@docs/en/observability/vaultak.mdx` around lines 1 - 6, The frontmatter for
the Vaultak Integration page currently sets icon="shield-check" while the
overview card uses icon="shield-halved"; update one of them so both use the same
icon string (e.g., change the frontmatter icon="shield-check" to
icon="shield-halved" or vice versa) to ensure consistent icons across the
overview card and the Vaultak page.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Nitpick comments:
In `@docs/en/observability/vaultak.mdx`:
- Around line 112-121: Update the example configuration block to include the
verbose parameter so readers see how to enable verbose logging; specifically add
verbose=True alongside the existing parameters (e.g., where the example
constructs the Vaultak client/config with api_key and agent_name) so the code
snippet demonstrates all five documented options including verbose.
- Around line 1-6: The frontmatter for the Vaultak Integration page currently
sets icon="shield-check" while the overview card uses icon="shield-halved";
update one of them so both use the same icon string (e.g., change the
frontmatter icon="shield-check" to icon="shield-halved" or vice versa) to ensure
consistent icons across the overview card and the Vaultak page.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 6e0fc09f-82cf-4f3d-8d68-6b49f1224318

📥 Commits

Reviewing files that changed from the base of the PR and between fca21b1 and cdac0c7.

📒 Files selected for processing (3)

• docs/docs.json
• docs/en/observability/overview.mdx
• docs/en/observability/vaultak.mdx

[samueloladji-beep]

@joaomdmoura @bhancockio — ready for review. This adds a Vaultak runtime security integration guide to the CrewAI observability docs.

[samueloladji-beep]

Hey @lorenzejay, all CI checks are passing — would appreciate a review when you get a chance. Thanks!