Liang Jiaqing
77d44c8cd0
update technical report
2026-04-19 10:03:32 +08:00
Liang Jiaqing
5d015d42c9
update technical report: language polish and table layout improvements
2026-04-18 19:58:56 +08:00
Liang Jiaqing
f8a380d4fd
fix: increase default timeout for non-stream mode (10/240s vs 5/30s)
2026-04-18 16:45:37 +08:00
Liang Jiaqing
2632478d96
fix: get_llm_name case-insensitive match & refactor to accept optional backend param
2026-04-18 14:46:05 +08:00
Liang Jiaqing
c529abce3c
fix: clarify start_long_term_update should only be called when task is completed (not mid-task)
2026-04-18 13:09:01 +08:00
Liang Jiaqing
11c41e7c4a
refine memory_cleanup_sop: 4 compression principles, RULES triage, anti-intuitive trigger words, layered placement
2026-04-18 12:33:41 +08:00
Liang Jiaqing
3c2d22b948
refactor: move chat bubble asset to frontends/ and rename to English
2026-04-18 11:13:38 +08:00
Keeper
5d0d4b7d27
add: right-click of changing skin for desktop pet ( #99 )
...
* fix: skin icon size
* fix: skin icon size
* fix: skin icon size
* fix: skin icon size
* fix: skin icon size
* fix: skin icon size
* add: right click logic of changing skin
2026-04-18 11:00:56 +08:00
LJQ
184948b772
Merge pull request #89 from mmgnosit-commits/fix/minimax-timeout-retry
...
fix(llmcore): 添加MiniMax超时错误码529支持重试机制
2026-04-18 10:59:16 +08:00
weijia
f49d69cf92
fix: 桌宠气泡 ( #102 )
...
* refactor: 精简desktop_pet_v2 — 内联_find_bubble_asset、删debug log、去重复RGBA检查
- 内联 _find_bubble_asset() 到 build_bubble_image (减少一层函数调用)
- 删除2处 /tmp/pet_toast_debug.log 写入
- 去除 Mac load_skin 中重复的 RGBA 转换检查
- 所有气泡核心逻辑(ImageOps.contain/非对称padding/alpha crop/tail扫描/定位公式)完整保留
- 785行 → 763行 (-22行)
* fix: 修复气泡文字碰边框问题，基于不透明区域计算文字padding
* feat: 添加自定义聊天气泡图片
* style: shift bubble text up 3px for better visual centering
2026-04-18 10:48:30 +08:00
Liang Jiaqing
ec34b7e1c0
Add technical report link to README
2026-04-18 00:13:48 +08:00
Liang Jiaqing
13b15f3255
Add technical report PDF
2026-04-18 00:09:36 +08:00
Liang Jiaqing
593eaa8cf7
Add Trendshift trending badge to README
2026-04-17 21:12:34 +08:00
Junghwan
adcfb8f730
fix(L4): preserve raw sessions skipped as recent(<2h) ( #90 )
...
compress_session.py Phase 4 builds to_del from both processed files
and every skipped file. Phase 1 marks files younger than 2h as
'recent(<2h)' (line 175) so their still-active writer is not
interrupted, but Phase 4 then deletes them anyway.
Filter the skipped-loop by reason so 'recent' files are preserved.
Processed files and other skip reasons (dup, compression error)
still proceed through deletion as before.
2026-04-17 17:35:50 +08:00
Jiaqing Liang
0e3e602121
fix: inline_eval exec() exception escaping crash; update schema description
2026-04-17 16:06:13 +08:00
Jiaqing Liang
47f106c8db
refactor: extract slash cmd handler; fix inline_eval param name; support extra_sys_prompt
2026-04-17 15:39:51 +08:00
宋明明
04b4818f87
fix(llmcore): 添加MiniMax超时错误码529支持重试机制
...
## 问题描述
MiniMax API在请求超时时返回HTTP 529状态码，该状态码未被包含在可重试状态码集合中， 导致即使配置了max_retries参数，重试机制也对MiniMax超时无效。
当前行为：MiniMax 529 → 不重试 → 直接报错
预期行为：MiniMax 529 → 触发重试逻辑 → max_retries生效 ## 解决方案
在llmcore.py的RETRYABLE集合中添加529状态码。 ## 关于其他改动的思考 最初的想法是：将 LLM API 重试的 HTTP 状态码从硬编码提取为可配置文件
assets/http_retry_codes.json，首次运行时自动生成默认配置
### 之前的初衷（配置文件方案）
- ✓ 不修改代码即可适配新的错误码
- ✓ 降低维护成本
### 现实考量（硬编码方案）
- 主流模型厂商就那几家：OpenAI、Claude、MiniMax、Kimi等 -
标准的超时错误码基本固化：408(timeout)、429(rate_limit)、5xx(server_error)
- MiniMax 529是特例但不频繁变化
- 硬编码更简洁直接，维护更清晰
## 受影响范围
- llmcore.py: _openai_stream() 函数的重试机制现在支持MiniMax 529错误
- 对应MiniMax API的超时场景现在能正确触发retry逻辑
2026-04-17 15:09:52 +08:00
Jiaqing Liang
f418963585
refactor: omit temperature from payload when default (1), avoid errors on reasoning models
2026-04-17 13:11:57 +08:00
Liang Jiaqing
9dbb3315da
docs: add English-only CONTRIBUTING.md
2026-04-17 10:46:09 +08:00
Liang Jiaqing
b1f72a202e
fix: clarify file_write usage — content before call
2026-04-17 00:04:12 +08:00
Liang Jiaqing
aa674655a8
refactor: unify docstrings, add /resume command, simplify error msg
2026-04-16 21:04:04 +08:00
LJQ
e7e560715e
Merge pull request #80 from shenhao-stu/main
...
docs: update mykey_template.py
2026-04-16 19:52:02 +08:00
LJQ
2c0146232a
Merge pull request #79 from ViviqwerAsd/docs/update-wechat-group-qrcodes
...
docs: 更新微信群二维码 (group 5/6/7)
2026-04-16 19:50:42 +08:00
Shen Hao
e4e84f2b43
docs: update mykey_template.py
2026-04-16 19:50:11 +08:00
Xinyi Wang
753bde713f
docs: update WeChat group QR codes (group 5/6/7) in README Support section
2026-04-16 19:45:57 +08:00
LJQ
f6deb83fe0
Merge pull request #78 from AspasZhang/feat/update-plan-verify-sop
...
feat: update plan_sop & verify_sop
2026-04-16 19:18:14 +08:00
Liang Jiaqing
c43c07487f
feat: add English version of global_mem_insight_template
2026-04-16 18:52:07 +08:00
Liang Jiaqing
7cadbd7403
feat: i18n support - auto-detect system language for zh/en prompts
2026-04-16 18:47:40 +08:00
张洲嘉
82b8535b7c
feat: update plan_sop & verify_sop
...
- plan_sop: improved planning workflow
- verify_sop: rewritten inspired by Claude Code verification agent design
- Added two failure mode anchoring (verification avoidance & seduced by first 80%)
- Anti-rationalization checklist
- BEFORE PASS/FAIL guardrails
- Compressed from 104 to 65 lines while increasing coverage
2026-04-16 18:42:03 +08:00
Liang Jiaqing
47bcb1b01a
fix: compatible with Streamlit 1.56+ iframe API change
2026-04-16 17:48:58 +08:00
Liang Jiaqing
9e18ce26dc
fix: empty file ZeroDivisionError in file_read; remove SiderLLMSession; init global_mem with header
2026-04-16 16:04:18 +08:00
Liang Jiaqing
8e2a96167e
docs: add token efficiency highlight to Core Features (EN+CN)
2026-04-16 15:47:41 +08:00
LJQ
315ce293df
Merge pull request #76 from kuishou68/fix/issue-75-scheduler-errors
...
fix: guard _parse_cooldown against malformed repeat values; close file handle with context manager
2026-04-16 13:22:48 +08:00
Liang Jiaqing
0591b68ff6
stapp: incremental freeze rendering for streaming segments
...
Replace full-redraw approach with frozen/live slot pattern:
- Completed segments are frozen into individual st.empty() slots
- Only the last (active) segment is re-rendered on each tick
- Reduces flicker and improves performance for long multi-turn conversations
2026-04-16 13:10:00 +08:00
Liang Jiaqing
3d7832a750
feat: multi code_run support, inline_eval logging & cwd fix, tool reinject upgrade
2026-04-16 12:56:17 +08:00
Cocoon-Break
fefd9b99cc
fix: guard _parse_cooldown against malformed every_ repeat; close file handle with context manager ( Closes #75 )
...
Signed-off-by: Cocoon-Break <54054995+kuishou68@users.noreply.github.com >
2026-04-16 11:31:35 +08:00
Liang Jiaqing
a3c15c54af
docs: add Star History chart to README
2026-04-16 07:58:48 +08:00
Liang Jiaqing
85480b16a6
docs: use wechat_group4.jpg to bypass CDN cache
2026-04-15 23:57:11 +08:00
Liang Jiaqing
4d6b8cd982
docs: update wechat group labels (Group 4 + Group 3)
2026-04-15 23:54:57 +08:00
Liang Jiaqing
39f90149fb
fix: change desktop pet port 51983->41983 (avoid Hyper-V reserved range) & update wechat group QR code
2026-04-15 23:50:48 +08:00
Liang Jiaqing
bdd2f3c2cd
docs: add Chinese illustrated getting-started guide
2026-04-15 23:22:45 +08:00
Liang Jiaqing
bd9e45f761
feat(file_read): fuzzy file suggestion on not-found with history dirs; sync L1 template
2026-04-15 20:58:29 +08:00
Liang Jiaqing
da8bba99a5
fix: batch_process missing raw_dir arg causing L4 archive to fail
2026-04-15 18:31:29 +08:00
Jiaqing Liang
e3a2526932
feat: add plan mode (enter/exit/completion check/verification gate/periodic hints)
2026-04-15 17:02:54 +08:00
LJQ
d482f30188
Merge pull request #69 from kagura-agent/feat/github-contribution-sop
...
feat: add GitHub contribution SOP (memory/github_contribution_sop.md)
2026-04-15 15:42:22 +08:00
Kagura
b2465df4e0
feat: add GitHub contribution SOP
2026-04-15 15:06:04 +08:00
wjl2023
bb0290f576
fix: normalize bubble text and adjust padding for better readability
2026-04-15 14:43:14 +08:00
Jiaqing Liang
16fe369f5a
preserve history across llm switch
2026-04-15 14:34:06 +08:00
Jiaqing Liang
d111db5454
fix prompt flow
2026-04-15 13:57:36 +08:00
Liang Jiaqing
273150a9ce
feat: publish memory_cleanup_sop
2026-04-15 09:26:46 +08:00
