zhayujieandcowagent ce3bbce2c0 fix: scope upload preview URLs to the selected agent ...

Uploads are stored per agent under <workspace>/tmp, but /upload returned a
preview_url without the agent, so /uploads/<name> resolved against the default
agent's tmp. Under any non-default agent the attachment 404'd and rendered as a
broken thumbnail.

Multipart uploads also ignored the agent entirely and always wrote to the
default workspace: the client sends it in the query string (a field present in
both query and body arrives as a list and breaks handlers), while the handler
read the body alone via rawinput("post").

Both upload routes now resolve the agent and name it in the URL they hand back,
matching what the voice ASR and TTS paths already did.

Co-authored-by: cowagent <cow@cowagent.ai>

2026-09-17 17:03:49 +08:00

..

conftest.py

refactor(web): gather the backend modules under channel/web/core/

2026-09-17 15:48:06 +08:00

test_agent_admin.py

fix: harden handling of references to deleted agents

2026-09-03 10:22:23 +08:00

test_agent_delegation.py

fix: delegate under the speaking guest's identity and let it reach the host

2026-09-13 21:21:26 +08:00

test_agent_initializer_routing.py

fix(agent): key cached runtime state by workspace, not by process

2026-08-02 17:33:06 +08:00

test_agent_model_defaults.py

feat(agents): address a teammate by name and keep the transcript attributed

2026-08-27 15:31:22 +08:00

test_agent_reasoning_effort.py

fix(models): treat qwen3.8-max/flash as hybrid thinking models

2026-09-16 12:33:53 +08:00

test_agent_registry.py

feat(channels): run multiple channel instances, each bound to its own agent

2026-09-01 19:38:39 +08:00

test_agent_routing.py

feat(channels): run multiple channel instances, each bound to its own agent

2026-09-01 19:38:39 +08:00

test_agent_steering.py

fix: surface stop_reason to the agent loop so a cut-off turn is distinguishable

2026-08-24 20:47:42 +08:00

test_agent_stream_content_blocks.py

fix: keep only the stream-level CoT fix, drop the channel-type gate

2026-08-16 14:57:39 +08:00

test_agent_trim_new_messages.py

fix(agent): persist AI reply when context is trimmed

2026-08-17 12:13:35 +08:00

test_agent_web_management.py

refactor(web): gather the backend modules under channel/web/core/

2026-09-17 15:48:06 +08:00

test_audio_convert.py

test(voice): cover the silk branch of any_to_mp3

2026-09-12 22:19:35 +08:00

test_baidu_translator.py

test(translate): cover the baidu retry-exhausted path

2026-09-12 17:05:10 +08:00

test_baidu_wenxin_image_create.py

test(baidu): pin the unsupported image-create reply

2026-09-12 14:22:10 +08:00

test_bash_background.py

feat(bash): run long commands in the background, and make stop actually stop

2026-07-27 20:01:09 +08:00

test_bash_config_propagation.py

feat(bash): run long commands in the background, and make stop actually stop

2026-07-27 20:01:09 +08:00

test_bash_exit_codes.py

fix(tools): three results that cost the model a turn for no reason

2026-07-28 12:12:28 +08:00

test_bash_output_encoding.py

fix(bash): write large-output temp file as UTF-8

2026-07-07 16:45:37 +08:00

test_bash_streaming.py

feat(bash): run long commands in the background, and make stop actually stop

2026-07-27 20:01:09 +08:00

test_bash_windows_guidance.py

fix(bash): keep bash wording on non-Windows, split Windows guidance

2026-08-20 10:34:00 +08:00

test_channel_double_start.py

fix(channel): stop a running channel before replacing it

2026-07-29 09:49:18 +08:00

test_channel_instances.py

refactor(scheduler): resolve live channel instance for delivery and use one global task store

2026-09-07 18:20:41 +08:00

test_chat_fallback_chain_migration.py

feat(chat): walk an ordered fallback chain instead of one backup model

2026-09-08 15:41:00 +00:00

test_chat_fallback_chain_walk.py

feat(chat): walk an ordered fallback chain instead of one backup model

2026-09-08 15:41:00 +00:00

test_chat_fallback_ratelimit.py

feat(chat): walk an ordered fallback chain instead of one backup model

2026-09-08 15:41:00 +00:00

test_chat_model_fallback.py

refactor(web): move ConfigHandler and ModelsHandler into channel/web/api/

2026-09-17 16:02:11 +08:00

test_chat_service_cancel.py

fix(web): close pre-registration cancel race

2026-08-20 14:33:33 +08:00

test_chrome_singleton_lock.py

fix(browser): stop deleting live Chrome profile locks

2026-07-28 12:36:31 +08:00

test_chunker_markdown.py

feat(memory): heading-aware chunking for memory/knowledge markdown files

2026-09-04 18:25:04 +00:00

test_chunker_version.py

feat(memory): track chunker version and suggest rebuild when stale

2026-09-05 01:23:13 +00:00

test_claude_effort.py

fix(tests): stop substituting stdlib re for the regex package

2026-08-26 16:23:13 +08:00

test_claude_thinking.py

test(claude): expect adaptive models to omit disabled thinking

2026-09-10 13:18:11 +08:00

test_cli_backup.py

feat(channels): run multiple channel instances, each bound to its own agent

2026-09-01 19:38:39 +08:00

test_cloud_channel_dispatch.py

feat(channels): run multiple channel instances, each bound to its own agent

2026-09-01 19:38:39 +08:00

test_config_subagent_toggle.py

refactor(web): move ConfigHandler and ModelsHandler into channel/web/api/

2026-09-17 16:02:11 +08:00

test_console_channel_manager_resolution.py

refactor(web): move the agents handlers into channel/web/api/

2026-09-17 16:12:31 +08:00

test_conversation_global_migration.py

feat: merge all agent conversations into one global index.db keyed by agent_id

2026-09-07 20:49:59 +08:00

test_conversation_runs.py

feat(scheduler): paginate and delete run records, richer run detail

2026-09-08 16:04:06 +08:00

test_cow_cli_tasks.py

feat: add cross-channel tasks command

2026-07-17 17:06:08 +08:00

test_custom_provider_handlers.py

feat(image): support custom generation providers

2026-08-10 16:03:04 +08:00

test_custom_provider.py

fix(models): resolve custom credentials from the routed provider, not global

2026-09-11 15:21:20 +00:00

test_custom_voice.py

feat: desktop voice input and custom providers for ASR/TTS

2026-08-16 02:35:34 +08:00

test_dashscope_provider.py

test(dashscope): assert plus-before-max order, not first place

2026-09-10 13:18:14 +08:00

test_dashscope_reasoning_effort.py

fix(models): treat qwen3.8-max/flash as hybrid thinking models

2026-09-16 12:33:53 +08:00

test_direct_addressing.py

refactor(web): gather the backend modules under channel/web/core/

2026-09-17 15:48:06 +08:00

test_doc_edit.py

refactor(web): move the remaining view handlers into channel/web/api/

2026-09-17 16:23:04 +08:00

test_edit_tool.py

refactor(edit): unify fuzzy uniqueness check with the fuzzy matcher

2026-07-08 11:13:30 +08:00

test_embedding_blank_input.py

fix(memory): keep blank text out of the embeddings API

2026-09-07 20:44:48 +08:00

test_evolution_safety.py

fix(evolution): serialize writes per workspace

2026-08-09 18:19:01 +08:00

test_evolution_trigger.py

fix: preserve evolution eligibility when concurrency is full

2026-08-05 16:49:52 +08:00

test_evolution.py

feat(evolution): give review agent full context, add knowledge signal, polish UX

2026-06-08 20:06:01 +08:00

test_fallback_provider_credentials.py

fix(models): resolve custom credentials from the routed provider, not global

2026-09-11 15:21:20 +00:00

test_fatal_error_classification.py

fix(agent): avoid context-overflow retry loop with output-token reserve

2026-08-20 16:07:35 +08:00

test_feishu_lark_install.py

fix(feishu): surface the SDK bundle download in the UI

2026-07-28 12:02:28 +08:00

test_feishu_markdown_images.py

feat: render remote images in Feishu cards

2026-07-17 17:15:37 +08:00

test_feishu_message_recall.py

feat(feishu): cancel recalled messages

2026-07-19 18:14:31 +08:00

test_feishu_progress_card.py

feat: simplify Feishu card config and update 2.1.4 docs

2026-07-19 13:15:47 +08:00

test_feishu_quoted_context.py

feat: include Feishu quoted message context

2026-07-17 17:09:48 +08:00

test_feishu_scheduler_card.py

feat(feishu): add scheduler action cards

2026-07-17 12:54:26 +08:00

test_feishu_static_card.py

feat: simplify Feishu card config and update 2.1.4 docs

2026-07-19 13:15:47 +08:00

test_image_generation_custom_provider.py

feat(image): support custom generation providers

2026-08-10 16:03:04 +08:00

test_image_generation_empty_b64.py

test(skills): cover empty b64_json/url handling in image results

2026-09-10 11:32:54 +08:00

test_image_generation_model_routing.py

feat(image-generation): support gpt-image-2.5-flare and gpt-image-2.5-sunburst

2026-09-11 15:32:03 +08:00

test_invariant_bash.py

fix(bash): narrow credential-file block to ~/.cow/.env only

2026-06-05 11:36:22 +05:30

test_keyword_attachment_urls.py

fix(keyword): detect attachment URLs by parsed path

2026-08-18 15:21:15 +08:00

test_knowledge_service.py

fix(knowledge): make graph links resolve and labels visible

2026-07-30 12:39:06 +08:00

test_knowledge_web.py

refactor(web): move the remaining view handlers into channel/web/api/

2026-09-17 16:23:04 +08:00

test_linkai_media_urls.py

test(linkai): cover media url classification

2026-09-13 10:45:02 +08:00

test_linkai_midjourney_send.py

test(linkai): cover the midjourney reply retry path

2026-09-13 12:44:05 +08:00

test_linkai_reasoning_effort.py

feat(reasoning): add LinkAI effort passthrough

2026-08-04 13:15:32 +08:00

test_linkai_voice_format.py

fix(voice): normalise the asr format guard in the linkai provider

2026-09-13 22:08:44 +08:00

test_mcp_session_recovery.py

fix: recover expired Streamable HTTP sessions

2026-09-01 20:48:42 +08:00

test_mcp_tool_prefix.py

feat: support MCP tool prefixes and Parallel Search setup

2026-09-04 15:37:23 -07:00

test_mcp_tool_retrieval.py

feat(mcp): add retrieval observability and fallback tests

2026-08-27 15:35:01 +08:00

test_memory_global_config.py

fix(agent): key cached runtime state by workspace, not by process

2026-08-02 17:33:06 +08:00

test_minimax_provider.py

feat(minimax): add MiniMax-M3, set as default, drop M2.5/M2.1/M2

2026-06-01 21:30:38 +08:00

test_model_catalog_api.py

refactor(web): move the session handlers into channel/web/api/

2026-09-17 16:15:13 +08:00

test_models_handler.py

refactor(web): move ConfigHandler and ModelsHandler into channel/web/api/

2026-09-17 16:02:11 +08:00

test_moonshot_reasoning_effort.py

fix(tests): stop substituting stdlib re for the regex package

2026-08-26 16:23:13 +08:00

test_multi_agent_runtime.py

feat(agents): address a teammate by name and keep the transcript attributed

2026-08-27 15:31:22 +08:00

test_multi_agent_state_isolation.py

feat: merge all agent conversations into one global index.db keyed by agent_id

2026-09-07 20:49:59 +08:00

test_openai_chat_api.py

refactor(web): move the workspace handlers into channel/web/api/

2026-09-17 16:19:16 +08:00

test_openai_http_attribution.py

feat(provider): attribute OrcaRouter gateway requests

2026-08-17 19:45:20 +08:00

test_own_cards.py

fix(subagent): drop the duplicate card and the skills a sub agent cannot use

2026-08-08 20:14:44 +08:00

test_parallel_tool_calls.py

feat(subagent): run parallel delegations at once and report each on its own

2026-08-04 21:04:31 +08:00

test_qianfan_provider.py

refactor(web): untangle ConfigHandler and ModelsHandler

2026-09-17 15:55:34 +08:00

test_qq_channel_session.py

fix(qq): auto-reconnect on silently-dead websocket connection

2026-09-12 14:41:47 +08:00

test_read_edit_improvements.py

fix(grep): stop announcing the denylist on every search

2026-07-27 17:26:01 +08:00

test_read_offset.py

fix(grep): stop announcing the denylist on every search

2026-07-27 17:26:01 +08:00

test_reasoning_capabilities.py

fix(models): treat qwen3.8-max/flash as hybrid thinking models

2026-09-16 12:33:53 +08:00

test_robustness_fixes.py

fix: avoid KeyError on cancel and infinite loop in image compression

2026-06-12 11:03:04 +08:00

test_run_lifecycle.py

feat(run): let a caller name a run and its parent

2026-08-26 16:19:12 +08:00

test_scheduled_task_empty_reply.py

fix: surface stop_reason to the agent loop so a cut-off turn is distinguishable

2026-08-24 20:47:42 +08:00

test_scheduler_channel_resolution.py

feat(scheduler): web-always-persist delivery, run preview/instance, newest-first ordering

2026-09-08 15:26:54 +08:00

test_scheduler_cross_channel_recipients.py

feat(web): redesign scheduled-task modal with channel-instance recipient picker

2026-09-07 18:21:32 +08:00

test_scheduler_run_now.py

fix: record manual scheduler trigger and thread it through the run

2026-09-08 15:21:36 +08:00

test_scheduler_run_records.py

feat(scheduler): paginate and delete run records, richer run detail

2026-09-08 16:04:06 +08:00

test_scheduler_silent.py

feat: refine scheduler silent mode to avoid mislabeling delivery tasks

2026-07-17 10:53:29 +08:00

test_scheduler_task_ordering.py

feat(scheduler): web-always-persist delivery, run preview/instance, newest-first ordering

2026-09-08 15:26:54 +08:00

test_scheduler_task_ownership.py

refactor(scheduler): resolve live channel instance for delivery and use one global task store

2026-09-07 18:20:41 +08:00

test_scheduler_web_update.py

refactor(web): move the scheduler handlers into channel/web/api/

2026-09-17 16:10:30 +08:00

test_search_files_tool.py

feat(tools): rename grep to search_files, add file-name search and a syntax gate

2026-07-27 18:40:20 +08:00

test_security_edit_env_bypass.py

fix(tools): close credential-guard bypass in the edit tool

2026-07-27 16:26:28 +08:00

test_security_read_env_bypass.py

test(read): add regression tests for credential-path bypass (#2913)

2026-07-06 15:30:13 +08:00

test_security_ssrf_browser_navigate.py

fix(browser): block link-local / cloud-metadata navigation (SSRF guard)

2026-06-19 20:15:27 +08:00

test_security_ssrf_path_traversal.py

fix(security): validate per-file paths in cloud skill install

2026-08-03 21:56:02 +08:00

test_security_ssrf_web_fetch.py

fix(tools): make web SSRF protection opt-in, disabled by default

2026-06-24 19:40:55 +08:00

test_session_model_catalog.py

refactor(web): move the session handlers into channel/web/api/

2026-09-17 16:15:13 +08:00

test_session_resurrect.py

fix: TLS trust in packaged builds, plus deleted-session and channel scan fixes

2026-07-28 17:03:32 +08:00

test_shared_db_recovery.py

fix(memory): detect interior page corruption on open again

2026-08-26 16:48:32 +08:00

test_skill_tar_security.py

fix(skill): harden tar archive extraction

2026-08-18 15:21:15 +08:00

test_ssl_ca_bundle.py

fix: TLS trust in packaged builds, plus deleted-session and channel scan fixes

2026-07-28 17:03:32 +08:00

test_state_dir.py

refactor(state): split workspace state into shared, per-agent and per-user roots

2026-08-11 10:27:43 +08:00

test_subagent_fallback_scope.py

feat(chat): walk an ordered fallback chain instead of one backup model

2026-09-08 15:41:00 +00:00

test_subagent_history.py

refactor(web): move the session handlers into channel/web/api/

2026-09-17 16:15:13 +08:00

test_subagent_run_records.py

feat(run): record spawned sub agent runs under their parent

2026-08-26 12:20:36 +08:00

test_subagent.py

fix(tests): match the delegation guards to the current prompt wording

2026-08-26 16:35:40 +08:00

test_syntax_gate.py

feat(tools): rename grep to search_files, add file-name search and a syntax gate

2026-07-27 18:40:20 +08:00

test_task_store_concurrency.py

fix(scheduler): prevent concurrent task updates from being lost

2026-08-18 15:21:15 +08:00

test_team_addressing.py

feat(team): bind a channel to an owner plus members for @mention and delegate

2026-09-01 19:38:48 +08:00

test_team_file.py

feat(channels): run multiple channel instances, each bound to its own agent

2026-09-01 19:38:39 +08:00

test_time_check.py

fix: accept end-of-day service stop time

2026-08-09 08:32:42 +08:00

test_tool_args_truncation.py

fix(agent): stop reporting a lost tool call as a missing parameter

2026-07-28 11:50:34 +08:00

test_tool_display.py

refactor: split console.js into a core/ and views/ script tree

2026-09-09 15:07:04 +08:00

test_tool_output_truncation.py

test(tools): pin the truncation reason reported for oversized output

2026-09-10 20:23:54 +08:00

test_tool_path_and_eval.py

fix(tools): three results that cost the model a turn for no reason

2026-07-28 12:12:28 +08:00

test_trajectory_eval.py

test: add agent trajectory evaluation

2026-09-03 17:06:04 +08:00

test_upload_agent_scope.py

fix: scope upload preview URLs to the selected agent

2026-09-17 17:03:49 +08:00

test_vector_backend.py

feat(memory): add pluggable vector backend

2026-08-10 15:51:47 +08:00

test_web_bind_failure.py

refactor(web): gather the backend modules under channel/web/core/

2026-09-17 15:48:06 +08:00

test_web_channel_disconnect.py

refactor(web): move the channels handlers into channel/web/api/

2026-09-17 16:07:25 +08:00

test_web_chat_content_type.py

refactor(web): move the last handlers out; web_channel.py is the URL table

2026-09-17 16:30:00 +08:00

test_web_console_assets.py

refactor(web): move the last handlers out; web_channel.py is the URL table

2026-09-17 16:30:00 +08:00

test_web_console_routing.py

refactor(web): move the last handlers out; web_channel.py is the URL table

2026-09-17 16:30:00 +08:00

test_web_console_update.py

refactor(web): move the remaining view handlers into channel/web/api/

2026-09-17 16:23:04 +08:00

test_web_search.py

refactor(web): move ConfigHandler and ModelsHandler into channel/web/api/

2026-09-17 16:02:11 +08:00

test_web_sse_cancel.py

fix(web): complete reliable SSE post-done lifecycle

2026-08-06 18:41:11 +08:00

test_web_sse_replay.py

refactor(web): move the last handlers out; web_channel.py is the URL table

2026-09-17 16:30:00 +08:00

test_weixin_credentials_path.py

feat(weixin): support multiple WeChat instances, each bound to its own agent

2026-09-01 20:22:43 +08:00

test_workspace_edit.py

refactor(web): move the workspace handlers into channel/web/api/

2026-09-17 16:19:16 +08:00

test_ws_client_compat.py

fix(qq): auto-reconnect on silently-dead websocket connection

2026-09-12 14:41:47 +08:00

test_youdao_translator.py

feat(translate): add Youdao as a new translation provider

2026-05-05 23:58:32 +08:00

test_zhipu_reasoning_effort.py

Fix GLM-5.3 always-thinking: force enabled thinking, add reasoning_effort SDK fallback, require zai-sdk>=0.2.3

2026-08-20 11:51:03 +08:00

trajectory_eval.py

test: add agent trajectory evaluation

2026-09-03 17:06:04 +08:00