Browse Source

Revert "docs: update README with v0.4.4 features (zh + en sync)"

This reverts commit 9a35302a2028637700e0f2f706926097cbfd9018.
Ma 7 months ago
parent
commit
234565f4f8
2 changed files with 6 additions and 46 deletions
  1. 3 23
      README.en.md
  2. 3 23
      README.md

+ 3 - 23
README.en.md

@@ -22,7 +22,7 @@ Open-source multi-agent system that autonomously writes, audits, and revises nov
 
 
 ## v0.4 Update
 ## v0.4 Update
 
 
-Spinoff writing + style cloning + post-write validator + audit-revise hardening + Writer two-phase split + multi-provider routing.
+Spinoff writing + style cloning + post-write validator + audit-revise hardening.
 
 
 ### Spinoff Writing
 ### Spinoff Writing
 
 
@@ -88,23 +88,6 @@ Real testing showed `rewrite` mode introduces 6x more AI markers than the origin
 - Re-audit temperature locked to 0 (deterministic pass/fail gating)
 - Re-audit temperature locked to 0 (deterministic pass/fail gating)
 - `polish` mode boundaries strengthened (no adding paragraphs, renaming entities, or changing causality)
 - `polish` mode boundaries strengthened (no adding paragraphs, renaming entities, or changing causality)
 
 
-### Writer Two-Phase Split
-
-Writer LLM calls are split into two phases: creative (temp 0.7, 8192 tokens) outputs prose only, settlement (temp 0.3, 8192 tokens) updates truth files only. Higher temperature for creativity, lower for factual accuracy. Chapter context compressed from last 3 chapters to last 1 + historical summaries.
-
-### Multi-Provider Routing
-
-Different agents can use different LLM providers (different baseUrl, API key, streaming settings). Use expensive models for Writer quality, cheap models for Auditor cost savings:
-
-```bash
-inkos config set-model auditor gemini-2.5-flash \
-  --base-url https://www.packyapi.com/v1 \
-  --api-key-env PACKYAPI_KEY \
-  --no-stream
-
-inkos config show-models   # View routing
-```
-
 ### Other v0.4 Changes
 ### Other v0.4 Changes
 
 
 - Audit dimensions expanded from 26 to 33 (+4 spinoff dims + dim 27 sensitive words + dim 32 reader expectation management + dim 33 outline adherence detection)
 - Audit dimensions expanded from 26 to 33 (+4 spinoff dims + dim 27 sensitive words + dim 32 reader expectation management + dim 33 outline adherence detection)
@@ -113,9 +96,6 @@ inkos config show-models   # View routing
 - Scheduler rewrite: AI-paced (15min cycles), parallel book processing, immediate retry, daily cap
 - Scheduler rewrite: AI-paced (15min cycles), parallel book processing, immediate retry, daily cap
 - New `spot-fix` revise mode (targeted repair)
 - New `spot-fix` revise mode (targeted repair)
 - `additionalAuditDimensions` in `book_rules.md` now supports name-string matching
 - `additionalAuditDimensions` in `book_rules.md` now supports name-string matching
-- Non-streaming LLM support (per-provider `stream: false`)
-- Error diagnostics: no longer swallows all exceptions as "inkos.json not found"
-- CLI version dynamically read from package.json
 
 
 ---
 ---
 
 
@@ -412,7 +392,7 @@ inkos up                          # Daemon mode
 | `inkos config set-global` | Set global LLM config (~/.inkos/.env) |
 | `inkos config set-global` | Set global LLM config (~/.inkos/.env) |
 | `inkos config show-global` | Show global config |
 | `inkos config show-global` | Show global config |
 | `inkos config set/show` | View/update project config |
 | `inkos config set/show` | View/update project config |
-| `inkos config set-model <agent> <model>` | Set model override for a specific agent (`--base-url`/`--provider`/`--api-key-env`/`--no-stream`) |
+| `inkos config set-model <agent> <model>` | Set model override for a specific agent |
 | `inkos config remove-model <agent>` | Remove agent model override (fall back to default) |
 | `inkos config remove-model <agent>` | Remove agent model override (fall back to default) |
 | `inkos config show-models` | Show current model routing |
 | `inkos config show-models` | Show current model routing |
 | `inkos doctor` | Diagnose setup issues (includes API connectivity test) |
 | `inkos doctor` | Diagnose setup issues (includes API connectivity test) |
@@ -494,7 +474,7 @@ TypeScript monorepo managed with pnpm workspaces.
 - [x] Style cloning (statistical fingerprint + LLM style guide + Writer injection)
 - [x] Style cloning (statistical fingerprint + LLM style guide + Writer injection)
 - [x] Post-write validator (11 hard rules + auto spot-fix)
 - [x] Post-write validator (11 hard rules + auto spot-fix)
 - [x] Audit-revise loop hardening (AI marker guard + temperature lock)
 - [x] Audit-revise loop hardening (AI marker guard + temperature lock)
-- [x] Multi-LLM provider (OpenAI + Anthropic + compatible endpoints + per-agent provider routing)
+- [x] Multi-LLM provider (OpenAI + Anthropic + compatible endpoints)
 - [x] AIGC detection + anti-detect rewrite pipeline
 - [x] AIGC detection + anti-detect rewrite pipeline
 - [x] Webhook notifications + smart scheduler (quality gates)
 - [x] Webhook notifications + smart scheduler (quality gates)
 - [x] Cross-chapter coherence (chapter summaries + subplot/emotion/character matrices)
 - [x] Cross-chapter coherence (chapter summaries + subplot/emotion/character matrices)

+ 3 - 23
README.md

@@ -22,7 +22,7 @@ Agent 写小说。写、审、改,全程接管。
 
 
 ## v0.4 更新
 ## v0.4 更新
 
 
-番外写作 + 文风仿写 + 写后验证器 + 审计闭环加固 + Writer 两阶段拆分 + 多提供商路由。
+番外写作 + 文风仿写 + 写后验证器 + 审计闭环加固。
 
 
 ### 番外写作(Spinoff)
 ### 番外写作(Spinoff)
 
 
@@ -88,23 +88,6 @@ inkos style import 参考小说.txt 吞天魔帝 --name 某作者  # 导入文
 - 再审温度锁 0(消除审计随机性,同一章不再出现 0-6 个 critical 的波动)
 - 再审温度锁 0(消除审计随机性,同一章不再出现 0-6 个 critical 的波动)
 - `polish` 模式加固边界(禁止增删段落、改人名、加新情节)
 - `polish` 模式加固边界(禁止增删段落、改人名、加新情节)
 
 
-### Writer 两阶段拆分
-
-Writer 单次 LLM 调用拆成两阶段:创作(temp 0.7, 8192 tokens)只输出正文,结算(temp 0.3, 8192 tokens)只更新 truth 文件。创作用高温度保持创造力,结算用低温度减少幻觉。章节上下文从最近 3 章全文压缩为最近 1 章 + 历史摘要。
-
-### 多提供商路由
-
-不同 agent 走不同 LLM 提供商(不同 baseUrl、API Key、流式设置),Writer 用贵模型保质量,Auditor 用便宜模型省成本:
-
-```bash
-inkos config set-model auditor gemini-2.5-flash \
-  --base-url https://www.packyapi.com/v1 \
-  --api-key-env PACKYAPI_KEY \
-  --no-stream
-
-inkos config show-models   # 查看路由
-```
-
 ### 其他 v0.4 变更
 ### 其他 v0.4 变更
 
 
 - 审计维度从 26 扩展到 33(+4 番外维度 + dim 27 敏感词 + dim 32 读者期待管理 + dim 33 大纲偏离检测)
 - 审计维度从 26 扩展到 33(+4 番外维度 + dim 27 敏感词 + dim 32 读者期待管理 + dim 33 大纲偏离检测)
@@ -113,9 +96,6 @@ inkos config show-models   # 查看路由
 - 修订者新增 `spot-fix` 模式(定点修复)
 - 修订者新增 `spot-fix` 模式(定点修复)
 - `book_rules.md` 的 `additionalAuditDimensions` 支持中文名称匹配
 - `book_rules.md` 的 `additionalAuditDimensions` 支持中文名称匹配
 - 全部 5 个题材激活 dim 24-26(支线停滞/弧线平坦/节奏单调)
 - 全部 5 个题材激活 dim 24-26(支线停滞/弧线平坦/节奏单调)
-- 非流式 LLM 支持(per-provider 可配 `stream: false`)
-- 错误诊断修复:不再把所有异常吞为 "inkos.json not found"
-- CLI 版本号从 package.json 动态读取
 
 
 ---
 ---
 
 
@@ -412,7 +392,7 @@ inkos up                       # 守护进程模式
 | `inkos config set-global` | 设置全局 LLM 配置(~/.inkos/.env) |
 | `inkos config set-global` | 设置全局 LLM 配置(~/.inkos/.env) |
 | `inkos config show-global` | 查看全局配置 |
 | `inkos config show-global` | 查看全局配置 |
 | `inkos config set/show` | 查看/更新项目配置 |
 | `inkos config set/show` | 查看/更新项目配置 |
-| `inkos config set-model <agent> <model>` | 为指定 agent 设置模型覆盖(`--base-url`/`--provider`/`--api-key-env`/`--no-stream`) |
+| `inkos config set-model <agent> <model>` | 为指定 agent 设置模型覆盖 |
 | `inkos config remove-model <agent>` | 移除 agent 模型覆盖(回退到默认) |
 | `inkos config remove-model <agent>` | 移除 agent 模型覆盖(回退到默认) |
 | `inkos config show-models` | 查看当前模型路由 |
 | `inkos config show-models` | 查看当前模型路由 |
 | `inkos doctor` | 诊断配置问题(含 API 连通性测试) |
 | `inkos doctor` | 诊断配置问题(含 API 连通性测试) |
@@ -508,7 +488,7 @@ TypeScript 单仓库,pnpm workspaces 管理。
 - [x] 题材自定义 + 单本书规则(genre CLI + book_rules.md)
 - [x] 题材自定义 + 单本书规则(genre CLI + book_rules.md)
 - [x] 33 维度连续性审计(含 AI 痕迹检测 + 番外维度 + 大纲偏离检测)
 - [x] 33 维度连续性审计(含 AI 痕迹检测 + 番外维度 + 大纲偏离检测)
 - [x] 去 AI 味铁律 + 文风指纹注入
 - [x] 去 AI 味铁律 + 文风指纹注入
-- [x] 多 LLM provider(OpenAI + Anthropic + 兼容接口 + per-agent 提供商路由)
+- [x] 多 LLM provider(OpenAI + Anthropic + 兼容接口)
 - [x] AIGC 检测 + 反检测改写管线
 - [x] AIGC 检测 + 反检测改写管线
 - [x] Webhook 通知 + 智能调度(质量门控)
 - [x] Webhook 通知 + 智能调度(质量门控)
 - [x] 跨章节连贯性(章节摘要 + 支线/情感/角色矩阵)
 - [x] 跨章节连贯性(章节摘要 + 支线/情感/角色矩阵)