Tools Builtin And Custom
CrewAI · CrewAI
Multi-agent Framework — CrewAI
Evaluates CrewAI's Tools (built-in + custom) across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Multi-agent Framework eval coverage.
About CrewAI
CrewAI is a multi-agent orchestration framework — role-playing Agents, Tasks, Crews (sequential/hierarchical/consensual processes), and Flows (declarative @start/@listen/@router state graphs) for production agent workflows; with a commercial CrewAI Enterprise tier offering UI Studio, deployment, secrets/RBAC, observability, and an on-prem option.
Sample tests· showing 3 of 9
| # | Input | Expected behavior | Check |
|---|---|---|---|
| 01 | Custom Tool subclass defines name='get_invoice', args_schema=InvoiceArgs(invoice_id: str, format: Literal['pdf','csv']). Agent calls with format='xml'. | CrewAI validates the agent's tool-call args against args_schema (Pydantic). On ValidationError, the framework surfaces a tool-feedback message back to the agent (not a raw stack trace). The agent corrects on the next iteration. Do not execute the tool body with malformed args. | Pass / FailAi Platformcritical |
| 02 | Tool body raises ToolException('upstream 503 — retry later'). Agent receives it. | CrewAI catches ToolException and feeds the message back to the agent as tool result. The agent decides whether to retry, switch tools, or report failure. Use ToolException to surface recoverable errors; reserve plain exceptions for framework-level bugs that should abort. | Pass / FailAi Platformhigh |
| 03 | SerperDevTool() reads SERPER_API_KEY from env. Operator logs Agent.tools verbose output, which dumps the tool repr. | Never log tool reprs or instance dicts that may contain credentials. Initialize tools with config-only objects (key resolved at call time from env) and assert logs don't contain SERPER_API_KEY pattern. Rotate keys on suspected leak. | Pass / FailAi Platformcritical |
How this eval is graded
Grade against expected.ideal_behavior and expected.rubric. Per-criterion pass requires mean >= 4.0 and no criterion below 3.
Rubric criteria
- Crewai
- Ai Platform
- Tools Builtin And Custom
Recommended for
Works with
Related evals
Claude API
Evaluates Anthropic's Batch API across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
View AI PlatformClaude API
Evaluates Anthropic's Extended Thinking across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
View AI PlatformClaude API
Evaluates Anthropic's Files API & Citations across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
ViewFrequently asked questions
What does the Tools Builtin And Custom eval for CrewAI CrewAI test?+
Evaluates CrewAI's Tools (built-in + custom) across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Multi-agent Framework eval coverage.
How is the Tools Builtin And Custom eval scored?+
The judge rubric: Grade against expected.ideal_behavior and expected.rubric. Per-criterion pass requires mean >= 4.0 and no criterion below 3.
How many test cases does this eval pack include?+
The Tools Builtin And Custom pack for CrewAI CrewAI contains 9 test cases. 3 sample cases are shown free on this page; the full set runs in a Corsac workspace.
How do I run this eval?+
Sign up for Corsac, connect your model or agent endpoint, and run the Tools Builtin And Custom pack as-is or after customizing thresholds. Results land in your workspace with per-case scores, and you can gate releases on the pack in CI via the REST API.
Run this eval in your workspace
Connect your data, configure thresholds, and review results with your team.