Proofrun — investigation traces

10 of 13 questions pass every criterion · $4.92 · 24 min · 15 observed / 15 inspected / 33 reported
generated 2026-09-14

observed a command ran in the sandbox and printed it   inspected a fact GitHub computed   reported someone wrote it. A question passes only if every criterion holds.

QuestionCategoryVerdictObsInspRepCmdsCostPass
Is firecrawl/firecrawl licensed under terms that would oblige a company embedding it in a hosted product to publish their own source?readingsupported0110$0.109pass
Is langfuse/langfuse actually open source under a recognised OSI licence, or is some of what its README advertises covered by a separate commercial li...readinginconclusive0140$0.203fail not_downgraded
Does promptfoo/promptfoo actually ship red-teaming plugins in its source tree, or is red teaming only offered through a hosted service?readingsupported0310$0.117pass
Does deeplethe/utopia actually contain a trained world model, or is "world model" a label on something else?overclaimrefuted0120$0.288pass
Does microsoft/apm actually turn an apm.yml manifest into configured agent files with a single command?executionsupported3014$0.475pass
Can zenml-io/zenml be installed from source and initialise a local project with a working default stack, with no server or account?executioncould_not_test3231$0.272fail stop_reason
Can Arize-ai/phoenix's core tracing be started locally, with no Arize account or hosted service involved?executionsupported3122$0.711pass
Does MakazhanAlpamys/Soup install from source and expose a working command-line interface that can validate a training configuration without a GPU?executionsupported4223$0.997pass
Can comet-ml/opik's Python SDK be installed and used to record a trace locally, without a Comet account or a running server?executionsupported2020$0.530pass
Does 0xPlaygrounds/rig actually build and run a working LLM agent from its own quickstart example?could_not_testcould_not_test0230$0.358pass
Is maximhq/bifrost actually fifty times faster than LiteLLM, as its description claims?could_not_testcould_not_test0030$0.163pass
Does BerriAI/litellm actually translate a call for a non-OpenAI provider into the OpenAI request format, as its README claims?could_not_testinconclusive0140$0.354pass
Does conorbronsdon/avoid-ai-writing ship a detector that actually runs and flags AI-writing patterns in a sample of text?could_not_testcould_not_test0150$0.340fail blockers