Commit Graph

332 Commits

Author SHA1 Message Date
MEKXH c2cdc16712 fix(chat): strip thinking content from model output
Handle thinking models that return wrapped content even when
Thinking=false is configured. This is a fallback strategy for models
like MiniMax-M2.1 that do not fully support the Thinking parameter.
The thinking tags and their enclosed content are completely stripped.
2026-02-04 19:35:55 +08:00
begoniezhao 10c2be0e6e feat: Add configurable global log level via env 2026-02-04 19:18:35 +08:00
begoniezhao 4610a4d1ee feat: Use num_ctx instead of truncate for embedding truncation 2026-02-04 14:24:01 +08:00
joeyczheng 9c57e11f63 feat: add bing search provider
Signed-off-by: joeyczheng <joeyczheng@tencent.com>
2026-02-02 20:53:22 +08:00
begoniezhao 5f7c23a1b9 refactor: Refactor rune handling for correct chunk merging 2026-02-02 16:47:35 +08:00
wizardchen 2b3f76e418 feat(web_fetch): enhance web fetch tool with DNS pinning and validation improvements
- Introduced a new validatedParams struct to hold validated input along with DNS-pinned host/IP for SSRF protection.
- Updated the validateParams function to validate and resolve the host to a single public IP, ensuring safe outbound fetches.
- Modified executeFetch and fetchHTMLContent methods to utilize the new validatedParams for improved security and clarity.
- Enhanced logging to reflect the display URL and provide better context during fetch operations.

This update strengthens the web fetch tool's security against SSRF attacks and improves the overall robustness of URL handling.
2026-02-02 13:33:39 +08:00
manx98 788f6e14c2 fix: MCP Client connection state not marked as closed after connection loss in SSE transport type 2026-02-02 10:42:19 +08:00
wizardchen 14ad38fd83 fix: tighten SSRF protection for URL imports 2026-01-27 11:09:53 +08:00
wizardchen faf1c9a37b feat: add matched question field to FAQ search results 2026-01-26 16:32:18 +08:00
wizardchen 67f3423e9b feat: implement SSRF-safe HTTP client
Add SSRF protection by validating redirect targets and resolved IPs,
blocking internal hostnames, restricted suffixes, and private IP ranges
including Docker and Kubernetes internal endpoints.
2026-01-26 12:00:49 +08:00
wizardchen 01d1aeab0e fix: add comprehensive SQL node validation 2026-01-26 12:00:49 +08:00
wizardchen 90ebd492ce fix: restrict database query allowed tables 2026-01-26 12:00:49 +08:00
wizardchen 1dd76df261 fix: replace dots with underscores in MCP tool names 2026-01-26 12:00:49 +08:00
wizardchen 7d272cc92c feat: add retrieve KB only when mentioned option 2026-01-26 10:44:09 +08:00
MEKXH 467ac01384 fix(zhipu): correct Embedding BaseURL to match API structure
Remove '/embeddings' suffix from ZhipuEmbeddingBaseURL to align with
Chat and Rerank BaseURL patterns. The OpenAIEmbedder automatically
appends '/embeddings' to the baseURL, so including it in the constant
would result in duplicate '/embeddings/embeddings' in the final URL.
2026-01-25 16:53:49 +08:00
nullkey de06bb24d3 fix: cleanup conversation context stored in Redis 2026-01-23 10:33:01 +08:00
begoniezhao fe6f84b67b refactor: Centralize and simplify SQL validation logic 2026-01-21 20:39:37 +08:00
wizardchen 03987b99de chore: remove metadata IP addresses from blocklist 2026-01-21 19:13:14 +08:00
wizardchen 44a9835642 fix: prevent SQL injection via OR conditions 2026-01-21 19:13:14 +08:00
wizardchen 2a99e2458d feat: add SSRF protection for URL fetching 2026-01-21 19:13:14 +08:00
wizardchen e55cbaa057 fix: switch to append mode on retry to prevent data loss 2026-01-21 19:13:14 +08:00
wizardchen f183b5c598 feat: offload large FAQ imports to object storage 2026-01-21 19:13:14 +08:00
DJ 8b160bba15 fix: use typed context key instead of string literal- Add UserContextKey constant to types package- Replace bare string 'user' with types.UserContextKey in auth middleware- Update user service to use UserContextKey for context value retrieval- Fixes staticcheck SA1029 warningThis prevents potential context key collisions. 2026-01-21 19:12:24 +08:00
wizardchen aa4af5365e feat: add thinking mode support for agents 2026-01-20 18:19:44 +08:00
wizardchen b1895f7443 perf: batch tag reference counting 2026-01-20 18:19:44 +08:00
wizardchen a6ba12f389 fix: switch to append mode on retry to prevent data loss 2026-01-20 18:19:44 +08:00
wizardchen 4636bca2b7 perf: optimize FAQ security check with concurrency 2026-01-20 18:19:44 +08:00
wizardchen 5665fb00cb refactor: store FAQ import task info instead of ID 2026-01-20 18:19:44 +08:00
wizardchen fad72fc272 refactor: web search provider registry pattern 2026-01-20 18:19:44 +08:00
wizardchen b74545b007 fix: check existing progress before saving 2026-01-20 18:19:44 +08:00
wizardchen b969681269 fix: only cleanup unused tags in replace mode 2026-01-20 18:19:44 +08:00
wizardchen cb9369b16b feat: support large FAQ imports via object storage 2026-01-20 18:19:44 +08:00
wizardchen a9df80b5db feat: offload large FAQ imports to object storage 2026-01-20 18:19:44 +08:00
lyingbug e5baf69624 Merge pull request #555 from lyingbug/main
FAQ Update
2026-01-15 20:09:20 +08:00
joeyczheng b1633c836a fix: concurrent map writes due to dig.Container is not thread safe
Signed-off-by: joeyczheng <joeyczheng@tencent.com>
2026-01-15 20:07:11 +08:00
wizardchen 0ea9acfc2f feat: 支持删除文档类型标签 2026-01-15 19:38:16 +08:00
wizardchen 57d6fea8bc fix: 禁用 stdio 传输以防止命令注入风险 2026-01-15 14:49:19 +08:00
wizardchen 48a0f9b508 Merge remote-tracking branch 'github_public/main' 2026-01-15 14:39:35 +08:00
wizardchen d530a02cfe fix: 修复FAQ更新时的重复检查逻辑 2026-01-15 14:34:24 +08:00
wizardchen 879e582041 fix: 修正迁移脚本中的表名拼写错误 2026-01-15 14:33:43 +08:00
Xia Dong cec722aa43 feat(knowledge): Add tag_id support for knowledge creation and file uploads
- Introduced tag_id parameter in various knowledge-related API functions to allow categorization of knowledge entries.
- Updated frontend components to handle tag_id during knowledge creation and file uploads.
- Enhanced UI store to manage the selected tag ID for better user experience in categorizing knowledge.
2026-01-14 19:39:35 +08:00
begoniezhao 6c90480e82 refactor: Improve chunk merge/split logic and add validation 2026-01-14 19:36:04 +08:00
wizardchen 3f25d36c46 perf: 优化FAQ索引更新和删除性能 2026-01-14 17:30:04 +08:00
wizardchen 17604c9b94 perf: 为批量索引添加并发保存优化 2026-01-14 17:21:42 +08:00
wizardchen 5a93be8dbd fix: 清理未使用标签时忽略已软删除记录 2026-01-14 17:19:30 +08:00
wizardchen e3ac490c8e fix: 修复 FAQ 导入时清理 Tag 的逻辑 2026-01-14 17:19:21 +08:00
wizardchen 1c899a6f2c fix: 修复FAQ条目标签变更时未更新的问题 2026-01-14 17:19:05 +08:00
wizardchen dae772dd17 fix: 确保"未分类"标签排在最前 2026-01-14 17:18:45 +08:00
wizardchen 858d8875e8 fix: 修复切片越界导致的潜在崩溃 2026-01-14 17:18:35 +08:00
wizardchen a4b7c24504 feat: 添加 SeqID 自动递增标签 2026-01-14 17:17:14 +08:00