All pull requests

Pull requests

• Add offline unit tests: coverage 42% → 59%

  #63·ipsorakis opened Oct 23, 2026·

  ·
  1

  

• feature: add AGENTS.md with Unit Testing Agent instructions

  #62·devin-ai-integration[bot] opened Oct 14, 2026Bot·

  ·
  1

  

• test: unit tests for dynamic_prompting, generic_agent and reproducibility_agent

  #61·devin-ai-integration[bot] opened Oct 14, 2026Bot·

  ·
  1

  

• test: unit tests for response_api + tracking (99% coverage, 5 xfail'd bugs)

  #60·devin-ai-integration[bot] opened Oct 14, 2026Bot·

  ·
  1

  

• test: unit tests for episode_to_html, agent_xray helpers, overlay_utils

  #59·devin-ai-integration[bot] opened Oct 14, 2026Bot·

  ·
  1

  

• test: unit tests for inspect_results and archive_studies

  #58·devin-ai-integration[bot] opened Oct 14, 2026Bot·

  ·
  4

• test: unit tests for study, reproducibility_util, args, view_dep_graph, multi_server

  #57·devin-ai-integration[bot] opened Oct 14, 2026Bot·

  ·
  1

  

• test: unit tests for agent_utils, most_basic, debug, visual and visualwebarena agents

  #56·devin-ai-integration[bot] opened Oct 14, 2026Bot·

  ·
  1

  

• test: unit tests for ToolUseAgent (blocks, StructuredDiscussion, get_action loop)

  #55·devin-ai-integration[bot] opened Oct 14, 2026Bot·

  ·
  1

  

• test: unit tests for experiments loop, exp_utils, launch_exp, ui_assistant

  #54·devin-ai-integration[bot] opened Oct 14, 2026Bot·

  ·
  1

  

• test: unit tests for llm_utils, prompt_templates and HF helpers

  #53·devin-ai-integration[bot] opened Oct 14, 2026Bot·

  ·
  1

  

• test: hermetic unit tests for chat_api and litellm_api (mocked clients)

  #52·devin-ai-integration[bot] opened Oct 14, 2026Bot·

  ·
  1

  

• test: unit tests for osworld preprocessing/action set, setup_benchmark, abstract_env

  #51·devin-ai-integration[bot] opened Oct 14, 2026Bot·

  ·
  1

  

• fix(deps): upgrade uv.lock for Critical/High advisories (13 Critical + 116 High -> 4 + 15); adapt agent_xray to gradio 6

  #50·vanessasalas-cog opened Oct 5, 2026·6 tasks·

  ·
  5

  

• test: mocked coverage for multi-turn tool-call formatting in response_api builders

  #49·giladsamuels-cognition opened Sep 30, 2026·

  ·
  1

  

• test: cover Study orchestration (make_study dispatch, cross-product, relaunch loop)

  #48·giladsamuels-cognition opened Sep 30, 2026·

  ·
  1

  

• test: cover cache-aware effective cost; fix Anthropic cache-read field name

  #47·giladsamuels-cognition opened Sep 30, 2026·

  ·
  4

  

• bug: guard OpenRouter pricing JSON parsing against malformed responses

  #46·devin-ai-integration[bot] opened Sep 23, 2026Bot·

  ·
  1

• feature: test StructuredDiscussion.flatten truncation and cache breakpoints

  #45·giladsamuels-cognition opened Sep 18, 2026·

  ·
  1

  

• feature: unit tests for generic_call_api_with_retries retry/backoff paths

  #44·giladsamuels-cognition opened Sep 18, 2026·

  ·
  1

  

• Cover VisualWebArena agent step/parse path with a non-pricy test

  #43·devin-ai-integration[bot] opened Sep 17, 2026Bot·

  ·
  1

• test: cover make_study dispatch, agent×benchmark enumeration and the relaunch loop

  #42·devin-ai-integration[bot] opened Sep 17, 2026Bot·

  ·
  1

• test: cover truncate_tokens start offset and no-op branches

  #41·devin-ai-integration[bot] opened Sep 17, 2026Bot·

  ·
  1

• test: cover Anthropic prompt-caching paths in response_api

  #40·devin-ai-integration[bot] opened Sep 17, 2026Bot·

  ·
  1

• test: cover ActionPrompt._parse_answer branches and Trunkater/fit_tokens edges

  #39·devin-ai-integration[bot] opened Sep 17, 2026Bot·

  ·
  1