Problemsby GlyphModels fail to return valid structured output
15 posts from 13 people11d activeSteady, +2 this week
Posts per day
The posts behind it
Agents window: how well 7 popular extensions work (PowerShell, Rainbow CSV, YAML, EditorConfig, Pretty TS Errors, Monokai Pro, Catppuccin)
I tried 7 popular Marketplace extensions in the Agents window to see whether they work there, especially when switching between sessions in different folders, and whether they are worth enabling there. The goal is to inform which extensions should be enabled for the Agents...
[BUG] After a checkpoint restore, a listener receives its trigger's Pydantic output as a dict, so `@human_feedback` listeners lose `.feedback`
### Description A flow method returns a Pydantic model, and its listener reads the model's attributes. The listener fails, and the flow is restored from its latest checkpoint and kicked off again. The resumed run calls the listener again, but now with a plain `dict` instead of...
[Bug]: Responses bridge duplicates tool-call arguments in /v1/chat/completions streaming (azure/responses)
### Description Streaming `/v1/chat/completions` requests that are bridged to the Responses API return each function call's arguments twice. The first `tool_calls` delta carries the complete arguments JSON, and the subsequent argument deltas then re-send the same arguments...
[BUG] CrewOutput.json raises bare IndexError when tasks_output is empty (#3185 was never fixed)
> Label: `llm-generated` - this report was produced with AI assistance (ZCode agent) and human-reviewed. Please apply the `llm-generated` label. ## Problem `CrewOutput.json` raises a bare `IndexError` when `tasks_output` is empty instead of the guided `ValueError` the code...
[EPIC] [Catalogsv2] Spark Iceberg catalog support
## Overview Tracks dbt v2 `catalogs.yml` support for selecting and using preconfigured Apache Spark catalogs, with an initial focus on Iceberg. This is a sub-epic of #16019. The original proof of concept is dbt-labs/fs#10485; it covered several catalog types in one PR and is...
feat: feat: Make Skill Toggles useful with native tool calling
...in skill management. ### Examples or References _No response_ ### Alternatives or Workarounds _No response_
[v2 Bug] v2 parser writes `columns: null` for model-level `check` constraints, which Core's v12 manifest schema rejects
...issues and could not find a duplicate ### Current Behavior Parsing with the v2 parser fails to produce a loadable manifest whenever a model declares a **model-level `check` constraint**: The v2 parser serializes the constraint's `columns` field as `null`: But in...
[BUG] VS Code extension: ScheduleWakeup/CronCreate stop being delivered to all sessions of a long-running extension host (silent, recovered only by full VS Code restart)
...not a factor - hooks, settings, cached feature flags: unchanged across the working and failing periods - same CLI from a terminal (2.1.293, TUI): wakeups delivered normally **Environment** - Windows 11 Pro 10.0.26300 - VS Code extension anthropic.claude-code 2.1.291 and 2.1.292...
[Bug]: Malformed Bedrock batch JSONL returns a server error instead of a client error
## Scope and evidence This is a suspected bug reported from patch and regression-test evidence for v1.100.0. A fresh standalone reproduction transcript is not yet attached; the steps below are the proposed reproduction, not a claim that they were freshly executed. ### Summary...
Validate params against each other in the trigger form
### Description allow a Dag to validate params against each other in the trigger form, so a bad combination blocks the trigger before a run is created. today each `Param` carries its own json schema, so the form can only check one field at a time. the [params...
Casting `variant_extract(v, 'k')` to JSON on a shredded VARIANT column parses the string instead of serializing it: `"true"` becomes the boolean `true`, `"x"` raises Malformed JSON
### What happens? On a VARIANT column whose row group has been **shredded** (at or above `variant_minimum_shredding_size`, default 30 000 rows), `variant_extract(v, 'k')::JSON` for a string-valued key returns the string's *contents parsed as JSON* rather than the string as a...
Python: [Python] as_agent_framework_tool drops parameter defaults (optionals become required)
## Description `KernelFunction.as_agent_framework_tool` builds an agent-framework `InputModel` but **drops parameter defaults**. After assigning a `Field(..., default=...)` for optional params, the next line unconditionally overwrites the same key with a no-default `Field`, so...
.Net: [Python] KernelJsonSchemaBuilder maps None to JSON "object" in multi-member unions
## Description `KernelJsonSchemaBuilder` maps `NoneType` to JSON type `"object"` in multi-member unions. A type like `int | str | None` becomes `anyOf: [integer, string, object]` instead of including `"null"`. This breaks function-calling / agent tool schemas for pydantic model...
FunctionTool(strict=True) loses strict across dump_component / load_component
### Describe the bug Docs require `strict=True` for tools used with structured output / `output_content_type`. `FunctionTool.dump_component` / `load_component` omit `strict` from the config, so a tool saved with `strict=True` comes back as `strict=False` after an apparently...
...outputs to gpt-5-mini made zero sense just because 5 was reasoning model and there was no way to disable the reasoning and it also did not have support for fine-tuning. So essentially I was not able to get nowhere close to the accuracy of previous model and it was slower, and...
Companies and products named
About this problem
Evidence
15 posts from 13 people in 11 places, about 2 a week over 45 days. Mostly on microsoft/semantic-kernel, BerriAI/litellm, dbt-labs/dbt. Tools named alongside: GitHub, DuckDB, Claude Code, VS Code.
Frustration Frustration 1 of 3· Seen on GitHub issues, Hacker News
How it was grouped
Posts that state a pain and match the "structured-output" rule. First post Sep 23, 2026, latest Nov 6, 2026. Corroborated: 3 or more posts from 2 or more people or places. Method
The same problem elsewhere
OpenAI fails to return valid structured output
12 posts from 12 people on GitHub issues, Hacker News.
Claude fails to return valid structured output
10 posts from 8 people on GitHub issues.
Codex fails to return valid structured output
6 posts from 3 people on GitHub issues.
Anthropic fails to return valid structured output
5 posts from 5 people on GitHub issues.
LlamaIndex fails to return valid structured output
5 posts from 5 people on GitHub issues.
Model Context Protocol fails to return valid structured output
5 posts from 5 people on GitHub issues, Reddit.
Ollama fails to return valid structured output
4 posts from 3 people on GitHub issues.
vLLM fails to return valid structured output
3 posts from 3 people on GitHub issues.
GitHub fails to return valid structured output
3 posts from 3 people on GitHub issues.
Gemini fails to return valid structured output
3 posts from 3 people on GitHub issues, Reddit.
Hugging Face fails to return valid structured output
2 posts from 2 people on Reddit, GitHub issues. Unverified.
LangChain fails to return valid structured output
2 posts from 2 people on GitHub issues. Unverified.
History