| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
docs: document /evolve — explanation, how-to and tutorial in both languages /evolve ships a 23-module Python sidecar, two search engines, four scoring modes, two agent tools, a design skill, 13 web components, 188 evolve.* message keys and four HTTP endpoints. Its documentation was one Chinese file, explanation/evolve-standalone.md, which no index linked and which answers a maintainer's question — how the two engines differ, and what it would take to run the sidecar as a standalone backend. In English the word "evolve" appeared in exactly one file, in passing. A researcher had no way to learn what a search is for, when scoring will actually discriminate, or how to start one. Add, in English and Chinese: - explanation/evolve.md — what a search is and the one property everything rests on, the four scoring modes with the case each suits, the two engines, the rollout/gate/test split, the discrimination probe, the frozen scorer, where candidates execute, and when not to run a search at all. - how-to/run-an-evolution-search.md — proposing a search, the four design checkpoints, shard sizing with the numbers that matter, choosing a seed, reading the panel, and a table of what to do when a run goes wrong. - tutorial/02-evolve-a-solution.md — one compression search a reader can complete, with what to look at at each checkpoint and how to tell a real improvement from a fitted one. The guidance is drawn from what the code and the skill already state rather than invented: the seed measurements (a cache task seeded with a working LRU climbing 0.2218 → 0.9396 against a compression task seeded with identity encoding finishing at 0.226), the gate/rollout/test sizing bands, and the upstream note on why the scorer is frozen. The seed guidance is framed around what the measurements actually show: the seed has to contain the mechanism the search will improve, in working form, because candidates edit what is in front of them and a from-scratch implementation fails far more often than an edit does. Headroom is a separate requirement — a starting point around 0.3–0.7 — rather than a reason to hand the search something deliberately poor. Screenshots come from the two merge requests that built the feature — !55 (PUCT) and !79 (OpenEvolve) — and are committed under docs/images/evolve/ rather than hot-linked. Also: link evolve-standalone.md from the Chinese explanation index, which takes the repository's orphan pages from three to two, and add a pointer at its top routing readers to the new user-facing pages; add the four evolve endpoints to both REST API references and the two agent tools to both built-in tool references. Refs openJiuwen-ai/sciencediscovery#55 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> | 16 天前 | |
docs(deploy): walk a Docker user from build to a signed-in browser Rewrite the Docker section of the deployment guide as numbered steps that end in a signed-in browser and a configured model, and add what a first deployment actually runs into: the data directory must exist before up or Docker creates it as root and the entry point refuses it; the sign-in URL in the log always names container port 4310, so a changed SCIENCE_AGENT_PUBLISH_PORT has to be substituted; the first start creates the starter Python environment in the background; the second instance needs its own pre-created data directory. Group the environment variables into orchestration, container and image layers so a key set in the wrong layer stops being a silent no-op, complete the Docker tables in the configuration reference (project name, image tag, host data directory, log and context settings, SSH config; drop the micromamba mirror URL, which the image does not need), and add a troubleshooting section for build, uid/gid, port, token, sandbox, proxy, host-side model, provisioning and health-check questions. Mirror the changes in the English guide and the deploy skill. | 14 天前 | |
feat(config)!: move the data directory to .sciencediscovery-data Both launchers wrote their runtime root to an undecorated directory: the repository launcher to <repo>/data, which was never gitignored and is easy to mistake for source, and the single-file launcher to ./science-discovery-data. Default both to .sciencediscovery-data, resolved from the repository root and from the launcher's working directory respectively. Changed in every place that declared a default: scripts/start-stack.sh, the API and Runner config loaders, the memory-graph logging config, .env.example, and the launcher CLI options and usage text. Existing installations are migrated once rather than orphaned. start-stack.sh moves an existing data/ directory, using the [compat] idiom the script already uses for the SCIENCE_AGENT_* to SCIENCE_DISCOVERY_* rename. The launcher now walks both former defaults, science-discovery-data before the older science-agent-data, so a host on either one is carried over; migrateLegacyDirectory already refuses to replace an existing target, so the newer wins and the older logs a skip. Neither path runs when the data directory is set explicitly. .dockerignore excludes the new name as well. That exclusion is what keeps bootstrap tokens and model-secrets.key out of the build context, so it has to move with the default; data/ stays excluded for un-migrated checkouts. Docker keeps its own /app/data bind mount. | 1 个月前 | |
docs: align quick-start, deployment, and MCP guides with token sign-in URL | 16 天前 | |
docs(evolve): name the single entry /evolve-design everywhere The composer already has one entry: the algorithm picker opens only for /evolve-design and no /evolve command is handled. The docs and code comments still described the feature as /evolve, which reads as a second entry. Rename those mentions to /evolve-design in the explanation, how-to, README and REST reference pages (en and zh) and in comments. Comments and prose only; the /api/evolve/... routes are untouched. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> | 13 天前 | |
docs: align quick-start, deployment, and MCP guides with token sign-in URL | 16 天前 | |
fix(web): clarify workspace disclosure state and localized labels | 22 天前 | |
!184 merge feat/node-native-web-providers into master feat: Node 原生 web 搜索/抓取，并改为先付费后免费的自动聚合 Created-by: wang_cheng_zhao Commit-by: wang_cheng_zhao Merged-by: wang_cheng_zhao Description: **What type of PR is this?** /kind feature --- **What does this PR do / why do we need it**: 在 [#181](https://gitcode.com/mindspore/ScienceAgent/pull/181) 把 agent 循环收进 Node 之后，本 PR 去掉剩余的 deer-flow **web 执行路径**，并把 web_search 改成自动聚合。 主要变化： - web_search / web_fetch 在 Node 内执行，不再 POST /internal/web/invoke，不再依赖 deerflow-harness。 - 删除 gateway HTTP 服务（loop/web sidecar）。gateway venv 仍可为随包 biomed / UniProt MCP 提供 Python 解释器。 - web_search 对外不再叫 DDGS：先试已配置密钥的付费引擎（Tavily → Exa → Brave API），再试已打开的免费引擎（DuckDuckGo HTML → 自接 Bing HTML → 免费 Brave HTML）。 - 设置页：付费勾选 + 密钥；免费引擎独立开关。旧 searchProvider / ddgsBackend 会迁移。 - WebBroker 的权限、缓存、CAS、审计、代理解析保持不变。 相关讨论见 [#87](https://gitcode.com/mindspore/ScienceAgent/issues/87)（移除 third_party/deer-flow 依赖的 RFC）。发布/首启里的 deer-flow 下载与 uv pip install 仍是后续清理项，本 PR 未改 launcher。 --- **Which issue(s) this PR fixes**: Related to #87 --- **Test Plan and Test result：What scenarios were tested, and what were the verification results（Function, performance, reliability, etc.）**： 被测 commit：6635c1f。独立检视结论：本次新增/改动文件无参考仓痕迹，聚合顺序与设置迁移正确，可合并。 | 范围 | 结果 | |---|---| | pnpm build | 通过 | | cd services/api && pnpm test | **450/450** | | cd apps/web && pnpm test | **350/350** | | cd services/gateway && python -m unittest discover -s tests | **16/16** | | web 专项单测（provider / broker / settings） | **26/26** | | 真实浏览器 E2E（API+Runner，无 gateway；真实出网） | **4/4 PASS** | E2E 场景： 1. 设置页无 DDGS；付费可配、免费三开关可见。 2. 关闭 Bing 后保存，重新打开仍保持。 3. 付费无 key 且免费全关：web_search 以 INVALID_INPUT 失败，助手不编造结果。 4. 默认免费聚合：DuckDuckGo 异常后 Bing 兜底成功，返回真实链接。 桌面 1280×720 设置区无横向溢出。未覆盖：付费厂商真实 API key 联调。 **E2E 截图**（被测 6635c1f）： 设置总览（付费 + 免费分区，无 DDGS）：  设置整页：  关闭免费 Bing，保存前：  重新打开后 Bing 仍关闭：  零候选：搜索失败可理解，未编造结果：  默认免费聚合成功（Bing 兜底）：  --- **Self-checklist**:（**请自检，在[ ]内打上x，我们将检视你的完成情况，否则会导致pr无法合入**） + - [x] **设计**：聚合顺序与设置模型已按产品要求落地；独立检视无阻塞项 + - [x] **测试**：单测与真实浏览器 E2E 已覆盖设置、失败提示与免费聚合成功路径 + - [x] **验证**：正文含命令级结果与 E2E 截图 + - [x] **接口**：Web 设置 schema 有迁移，旧字段读入后转换为新结构 + - [x] **文档**：仓内 web-tools / architecture 等已同步，不涉及官网资料仓 See merge request: mindspore/ScienceAgent!184 | 1 个月前 |
| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
| 16 天前 | ||
| 14 天前 | ||
| 1 个月前 | ||
| 16 天前 | ||
| 13 天前 | ||
| 16 天前 | ||
| 22 天前 | ||
| 1 个月前 |