# Inside VAKRA: Reasoning, Tool Use, and Failure Modes of Agents

DevFeed: [Inside VAKRA: Reasoning, Tool Use, and Failure Modes of Agents](<https://devfeed.tech/articles/inside-vakra-reasoning-tool-use-and-failure-modes-of-agents-7270.md>)

Original publisher: [Read original article](<https://huggingface.co/blog/ibm-research/vakra-benchmark-analysis>)

Author: Ankita Naik; Danish; Ben; Anupama Murthi; Praveen Venkateswaran; Siyu; Ayhan Sebin

Published: 2026-04-15T12:07:25Z

Content type: article

Language: en

Sources: [Hugging Face - Blog](<https://devfeed.tech/sources/hugging-face-blog.md>)

Topics: [Benchmark](<https://devfeed.tech/topics/benchmark.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [dataset](<https://devfeed.tech/topics/dataset.md>), [API](<https://devfeed.tech/topics/api.md>), [Databases](<https://devfeed.tech/topics/databases.md>), [Traces](<https://devfeed.tech/topics/traces.md>), [Model Context Protocol](<https://devfeed.tech/topics/model-context-protocol.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [api](<https://devfeed.tech/tags/api.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [database](<https://devfeed.tech/tags/database.md>), [mcp](<https://devfeed.tech/tags/mcp.md>), [tool](<https://devfeed.tech/tags/tool.md>), [traces](<https://devfeed.tech/tags/traces.md>)

## AI overview

VAKRA is a tool-grounded, executable benchmark for evaluating how AI agents reason and act in enterprise-like environments. It tests compositional reasoning across APIs and documents through multi-step workflows, execution traces, tool use, retrieval, and interactions with locally hosted APIs and databases.

## Source excerpt

VAKRA Dataset | LeaderBoard | Release Blog | GitHub | Submit to Leaderboard We recently introduced VAKRA, a tool-grounded, executable benchmark for evaluating how well AI agents reason and act in enterprise-like environments. Unlike traditional benchmarks that test isolated skills, VAKRA measures compositional reasoning across APIs and documents, using full execution traces to assess whether agents can reliably complete multi-step workflows.