Releases·Tool Release
A small CLI for checking nested tool calls, streaming, and the next turn
A developer released toolcall-check, a small Python CLI for testing chat completions compatible endpoints. It probes forced function calls, streaming, nested argument integrity, and multi-turn tool result round trips, producing a private HTML report with sanitized failure traces. The author is seeking feedback, noting the tool has not yet been tested against remote models.
9 engagement1 sourceSat, Oct 3, 2026, 11:20 AM
AI Overview
- toolcall-check, a Python CLI, tests nested argument JSON type integrity, streamed fragment reconstruction, a two-turn round trip, and sanitized-trace HTML reports
- The demo runs against a synthetic local fixture over real HTTP, not remote models; author requests feedback on fixed probes and the strict [DONE] requirement
- The author points to CompatCanary as already covering broader compatibility scans with forced calls, streaming, and structured output
Entities
Related
Products·building
Agent regression testing tool that turns agent failures into repeatable tests
A tool for regression testing LLM agents: it replays workflows in isolation and checks whether the agent called the right tools with correct arguments and completed the task. This addresses the problem that agents can produce convincing final responses even when tool calls fail or APIs break mid-workflow. It is aimed at builders who need to verify agent behavior before shipping prompt or code changes.
A tool for regression testing LLM agents: it replays workflows in isolation and checks whether the agent called the right tools with correct arguments and completed the task. This addresses the problem that agents can produce convincing final responses even when tool calls fail or APIs break mid-workflow. It is aimed at builders who need to verify agent behavior before shipping prompt or code changes.
15 engagement1 sourceredditTue, Sep 8, 2026, 11:59 AM
Releases·Tool Release
InfernoSIM v4.0 open-source failure simulator for AI agents released
The author of InfernoSIM released v4.0, a local reliability-testing tool for tool-using agents. It records sanitized model and tool traffic, replays it deterministically, injects failures, and verifies actual agent behavior. The author seeks engineers to test it on real stacks.
The author of InfernoSIM released v4.0, a local reliability-testing tool for tool-using agents. It records sanitized model and tool traffic, replays it deterministically, injects failures, and verifies actual agent behavior. The author seeks engineers to test it on real stacks.
2 engagement1 sourceredditMon, Aug 24, 2026, 01:54 PM
News·Incident
Sonnet 5.5 Streaming Broken on all providers when tool calls are present
A Reddit post reports Claude Sonnet 5.5 fails to stream when a request ends in a tool_result: all text is held until the message completes, then arrives in one burst. The issue appears on the Anthropic API, Bedrock, and OpenRouter, while Claude 4.6 streams normally. A reproduction with data is filed as GitHub issue #1960 on anthropic-sdk-python.
A Reddit post reports Claude Sonnet 5.5 fails to stream when a request ends in a tool_result: all text is held until the message completes, then arrives in one burst. The issue appears on the Anthropic API, Bedrock, and OpenRouter, while Claude 4.6 streams normally. A reproduction with data is filed as GitHub issue #1960 on anthropic-sdk-python.
4 engagement1 sourceredditWed, Sep 30, 2026, 07:26 AM
Releases·Tool Release
AgentSnap snapshot-testing library for LLM agents released as open source
A developer released AgentSnap, an MIT-licensed snapshot-testing library that records LLM agent calls as golden snapshots and diffs future runs to catch silent regressions. The tool compares tool-call sequences, arguments, model-requested tool calls, and semantic content of responses.
A developer released AgentSnap, an MIT-licensed snapshot-testing library that records LLM agent calls as golden snapshots and diffs future runs to catch silent regressions. The tool compares tool-call sequences, arguments, model-requested tool calls, and semantic content of responses.
31 engagement2 sourcesreddit, githubWed, Jul 29, 2026, 08:19 PM
Products·launched
Snapshot testing tool for multi-agent LLM pipelines to catch regressions
An open-source tool that performs pytest-style snapshot testing for multi-agent LLM pipelines. It captures full traces of agent prompts, I/O, and tool calls, then pins them behind content hashes to detect behavioral regressions when prompts or models are changed. Designed for teams running complex agent pipelines who need to catch subtle degradations that unit tests miss.
An open-source tool that performs pytest-style snapshot testing for multi-agent LLM pipelines. It captures full traces of agent prompts, I/O, and tool calls, then pins them behind content hashes to detect behavioral regressions when prompts or models are changed. Designed for teams running complex agent pipelines who need to catch subtle degradations that unit tests miss.
1 engagement1 sourceredditFri, Jul 17, 2026, 03:37 PM