Add zero-code examples for CrewAI and Anthropic Claude agents

[YaMachinu]

Context

Currently, agentevals ships zero-code examples and E2E tests for LangChain and Strands only. Two high-value frameworks are missing examples that would significantly broaden the project's market reach:

1. CrewAI — one of the most popular multi-agent Python frameworks, with a large community
2. 2. Anthropic Claude agents — critical alignment with ecosystem and the product's own tooling (the MCP server targets Claude Code users)

What's Needed

CrewAI Example (examples/crewai_example.py)

CrewAI uses OpenAI-compatible APIs and can be instrumented with opentelemetry-instrumentation-openai-v2, meaning it likely works with the existing GenAI semconv logs-based path. A zero-code example would look similar to the LangChain one:

OPENAI_API_KEY=... OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4318 \
  opentelemetry-instrument python crewai_example.py

Anthropic Claude Agent Example (examples/anthropic_example.py)

The Anthropic SDK can be instrumented with opentelemetry-instrumentation-anthropic (community package) or via the opentelemetry-instrumentation-openai-v2 with Anthropic's OpenAI-compatible endpoint. An example using the Anthropic SDK directly with OTel instrumentation would validate this path.

Why This Matters

• CrewAI has massive adoption and would drive organic discovery
• • An Anthropic/Claude example is essential given on strategic focus on the Claude ecosystem
• • • Both examples would validate the GenAI semconv code path with non-OpenAI models

[ossama-ferjani]

Hello @krisztianfekete can I take this issue?

[krisztianfekete]

Sure, but please file an individual PR for each, thanks!

[ossama-ferjani]

@krisztianfekete here's the fix. P.S please test the Anthropic changes( not tested I don't have a key just checked syntax and implementation) the crewai is tested.