kai/deep_research
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

v0.20.5 add antigravity clean workspace

Browse Source
This commit is contained in:
Deep Research System
2026-05-07 20:47:33 +08:00
parent a0f1144f6c
commit bb2c91754b
27 changed files with 3501 additions and 67 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/antigravity-clean-workspace.md
+90
View File
@@ -0,0 +1,90 @@
# Antigravity Clean Workspace
Use a clean workspace for Antigravity. Do not open the full multi-platform repository unless you are debugging adapters.
The full repository contains Codex, OpenCode, Claude Code, Gemini CLI, archived projects, and legacy adapter files. Gemini models may read those files and follow stale or incompatible instructions. The clean workspace exposes only the Antigravity instruction stack plus optional Python utilities.
## Required Directory Logic
Antigravity workspace files live under `.agent/`:
- `.agent/agents.md` defines roles only.
- `.agent/rules/*.md` defines persistent workspace constraints.
- `.agent/skills/*/SKILL.md` defines reusable procedures.
- `.agent/workflows/*.md` defines slash-style workflows.
Legacy or other-platform directories are not part of Antigravity native execution:
- `.agents/` is legacy compatibility for this repository's older adapter layout.
- `.codex/`, `.opencode/`, `.claude/`, and `.gemini/` are for other tools.
- Antigravity native runs must ignore those directories unless the user explicitly asks to inspect them.
## Sparse Checkout From Gitea
Use this when preparing an Antigravity-only checkout:
```bash
git clone --filter=blob:none --no-checkout http://192.168.50.45:3000/kai/deep_research.git deep_research_antigravity
cd deep_research_antigravity
git sparse-checkout init --no-cone
git sparse-checkout set \
AGENTS.md \
GEMINI.md \
README.md \
.agent \
docs/antigravity-clean-workspace.md \
docs/platform-adapters.md \
configs \
scripts/dr.py \
scripts/runtime \
scripts/reporting \
scripts/deploy_adapters.py \
projects/.gitkeep
git checkout codex/v0.20-skill-driven-python-core
```
This checkout intentionally excludes `.codex`, `.opencode`, `.claude`, `.gemini`, `.agents`, `archive`, historical `projects/**`, and bundled platform templates.
## Export From A Full Checkout
If you already have a full checkout, export a clean Antigravity workspace:
```bash
python scripts/export_antigravity_workspace.py --target /path/to/deep_research_antigravity --force
```
Open `/path/to/deep_research_antigravity` in Antigravity, not the full repository.
## Startup Prompt
Paste this as the first message in Antigravity:
```text
You are running Antigravity native Deep Research.
First, read only these files:
1. AGENTS.md
2. GEMINI.md
3. .agent/agents.md
4. .agent/rules/deep-research-antigravity.md
5. .agent/skills/antigravity-surface-adapter/SKILL.md
6. .agent/skills/method-selection/SKILL.md
7. .agent/skills/research-quality-gates/SKILL.md
8. .agent/workflows/deep-research-native.md
Ignore .codex, .opencode, .claude, .gemini, and .agents if present.
Use Gemini 3 Flash only as Surface Manager. Do not write report content or invent completed searches. Ask me to switch models at the workflow gates.
Research topic: <topic>
Slug: <slug>
Report purpose: <purpose>
Target reader: <reader>
Decision supported by the report: <decision>
Start with Step 1 of .agent/workflows/deep-research-native.md. Stop after Phase 1 artifacts are created and ask for approval.
```
## Failure Stop Rule
If the model claims it searched, cleared old files, wrote a chapter, ran a pipeline, or verified evidence without producing the corresponding artifact paths and receipts, stop the run. Restart with the startup prompt above and require it to show the loaded file list first.
docs/platform-adapters.md
+23 -14
View File
@@ -1,6 +1,6 @@
# v0.20 Platform Adapters
> v0.20 的默认核心入口是 Python core：`scripts/dr.py`、`scripts/runtime/**`、`configs/models.yaml`、`.agents/skills`。Antigravity 例外：它使用 `.agents/skills` 与 `.agents/rules` 走 native Deep Research，让 Antigravity 模型配额承担主要研究工作。
> v0.20 的默认核心入口是 Python core：`scripts/dr.py`、`scripts/runtime/**`、`configs/models.yaml`、`.agent/skills`。Antigravity 例外：它使用 `.agent/skills` 与 `.agent/rules` 走 native Deep Research，让 Antigravity 模型配额承担主要研究工作。
## Shared Rule
@@ -132,15 +132,23 @@ gemini
官方资料显示，Antigravity 是 agent-first IDE：主 Agent 可跨 editor、terminal、browser 工作；浏览器操作由 browser subagent 执行；Agent Manager 可管理多个任务流和 artifacts。基于 Codex 的 skill-driven 经验，本项目在 Antigravity 中不再默认把研究主流程交给 Python model workers，而是让 Antigravity 用自己的模型配额按 phase 执行研究。
本项目已提供 Antigravity 入口：
本项目已提供 Antigravity 入口。注意：Antigravity 入口使用 `.agent/` 单数目录；`.agents/` 是旧版兼容目录，不作为 Antigravity native 的加载目标。
- `.agents/agents.md`：Antigravity 角色定义，只定义 Research Manager / Strategist / Analyst / Reviewer / Final Editor 等角色。
- `.agents/skills/antigravity-surface-adapter/SKILL.md`：Antigravity 专用执行说明。
- `.agents/skills/method-selection/SKILL.md`：按场景选择研究方法，避免默认套用麦肯锡/MECE。
- `.agents/skills/research-quality-gates/SKILL.md`：把 search receipt、source access、claims ledger、delta retrieve、continuation state 和 final fact check 做成阶段质量门。
- `.agents/rules/deep-research-antigravity.md`：workspace rule，约束其优先使用 Antigravity 模型配额，避免主流程落回 ZenMux。
- `.agents/workflows/deep-research-native.md`：可按 slash workflow 执行的四阶段研究流程与反幻觉 gate。
- `.agents/skills/*`：Antigravity 官方默认 skill 目录，直接复用本项目 canonical skill registry。
- `.agent/agents.md`：Antigravity 角色定义，只定义 Research Manager / Strategist / Analyst / Reviewer / Final Editor 等角色。
- `.agent/skills/antigravity-surface-adapter/SKILL.md`：Antigravity 专用执行说明。
- `.agent/skills/method-selection/SKILL.md`：按场景选择研究方法，避免默认套用麦肯锡/MECE。
- `.agent/skills/research-quality-gates/SKILL.md`：把 search receipt、source access、claims ledger、delta retrieve、continuation state 和 final fact check 做成阶段质量门。
- `.agent/rules/deep-research-antigravity.md`：workspace rule，约束其优先使用 Antigravity 模型配额，避免主流程落回 ZenMux。
- `.agent/workflows/deep-research-native.md`：可按 slash workflow 执行的四阶段研究流程与反幻觉 gate。
- `.agent/skills/*`：Antigravity 官方默认 skill 目录，直接复用本项目 canonical skill registry。
推荐方式是先创建 clean workspace，只让 Antigravity 看到必要文件：
```bash
python scripts/export_antigravity_workspace.py --target /path/to/deep_research_antigravity --force
```
或使用 sparse checkout，见 `docs/antigravity-clean-workspace.md`。
部署到另一个 Antigravity workspace：
@@ -149,11 +157,11 @@ uv run python scripts/deploy_adapters.py antigravity --target /path/to/workspace
uv run python scripts/deploy_adapters.py antigravity --target /path/to/workspace
```
该命令只写 workspace-local `.agents/agents.md`、`.agents/skills`、`.agents/rules` 和 `.agents/workflows`，不修改全局 Antigravity/Gemini 设置；已有文件默认跳过，`--force` 才会备份为 `.bak` 后覆盖。
该命令只写 workspace-local `.agent/agents.md`、`.agent/skills`、`.agent/rules` 和 `.agent/workflows`，不修改全局 Antigravity/Gemini 设置；已有文件默认跳过，`--force` 才会备份为 `.bak` 后覆盖。
本项目调用方式：
1. 在 Antigravity 打开仓库根目录。
1. 在 Antigravity 打开 clean workspace 根目录，不要打开完整多平台仓库。
2. 在 Customizations / Rules 中确认 workspace rule 已启用；若未自动启用，手动 at mention `deep-research-antigravity` 或直接引用该文件。
3. 选择模型：
@@ -168,7 +176,8 @@ uv run python scripts/deploy_adapters.py antigravity --target /path/to/workspace
4. 给 Agent Manager 一个明确任务：
```text
请加载 .agents/agents.md、.agents/rules/deep-research-antigravity.md、.agents/skills/antigravity-surface-adapter/SKILL.md、.agents/skills/research-quality-gates/SKILL.md 和 AGENTS.md。
请加载 .agent/agents.md、.agent/rules/deep-research-antigravity.md、.agent/skills/antigravity-surface-adapter/SKILL.md、.agent/skills/research-quality-gates/SKILL.md 和 AGENTS.md。
忽略 .codex、.opencode、.claude、.gemini 和 legacy .agents。
用 Gemini 3 Flash 做流程管理，按 skill 要求在需要时提示我切换模型。
研究主题："ADC 全球竞争格局"，slug: adc-global-landscape，method: mckinsey_market。
请先创建/确认项目目录，然后进入 Phase 0-1，并在生成 phase1/framework.md 和 research_brief 后暂停。
@@ -182,7 +191,7 @@ uv run python scripts/deploy_adapters.py antigravity --target /path/to/workspace
- Python core 可用于 `init/status/skills validate/finalize/render` 等辅助动作。
- 若本机安装了 `paperfoot/search-cli` 的 `search` 命令，Antigravity Phase 0-2 可优先用 `search agent-info`、`search ... --json` 做统一检索入口。
- Phase 2 不只产出章节草稿；必须维护 `search_log.jsonl`、`sources.jsonl`、`claims_ledger.jsonl`、`coverage_matrix.md` 和 `unsupported_claims.md`，否则 Phase 3/4 不得声称已验证。
- 不建议依赖 SOUL.md：它不是 Antigravity 官方 Deep Research 约束入口。本项目用 `AGENTS.md`、`.agents/agents.md`、`.agents/rules`、`.agents/skills` 和 `.agents/workflows` 分层约束。
- 不建议依赖 SOUL.md：它不是 Antigravity 官方 Deep Research 约束入口。本项目用 `AGENTS.md`、`.agent/agents.md`、`.agent/rules`、`.agent/skills` 和 `.agent/workflows` 分层约束。
## Model Strategy Across Platforms
@@ -192,7 +201,7 @@ uv run python scripts/deploy_adapters.py antigravity --target /path/to/workspace
| Codex | GPT 系列用于代码、测试、schema、review | `configs/models.yaml` |
| Claude Code | Opus/Sonnet 用于访谈、中文风格、结构讨论 | `configs/models.yaml` |
| Gemini CLI | Gemini 用于长上下文、多模态、框架审阅 | `configs/models.yaml` |
| Antigravity | Gemini Flash/Opus/Gemini Pro 分 phase 执行研究 | `.agents/skills` + Antigravity model selector |
| Antigravity | Gemini Flash/Opus/Gemini Pro 分 phase 执行研究 | `.agent/skills` + Antigravity model selector |
核心原则：除 Antigravity 外，平台模型负责“怎么帮用户操作项目”，ZenMux/Python role 模型负责“研究任务用哪个模型执行”。Antigravity 专门走 native 模式，以消耗 Antigravity 自身模型配额并降低 Python-core 碎片化风险。