billion-context-pi 0.1.68-pr.411.207 → 0.1.68

package/README.md

@@ -10,16 +10,6 @@ The model decides <em>when</em> and <em>what</em> to compress — not a hard lim
 
 ---
 
-## 📄 Paper / Preprint
-
-- **[Model-Driven Incremental Hierarchical Compression: Training-Free Multi-Generational Context Management for Long-Lived Coding Agents](./paper/model-driven-incremental-hierarchical-compression-training-free-multi-generational-context-management-for-long-lived-coding-agents.md)** (English, v0.2)
-
-> 📝 **The paper itself is open-sourced under the MIT License as part of the codebase (`paper/`). It is a living document — anyone may edit it; improvements are welcome via pull request.**
-
-A production-scale longitudinal study: 4.5 months, three hosts, 174,327 model calls, 18.76B cumulative input tokens (~24.7B across all hosts), zero window violations on 204,800-token models, marathon sessions of 8,584–12,049 calls.
-
----
-
 <p align="center">
 <a href="https://www.npmjs.com/package/billion-context-pi"><img src="https://img.shields.io/npm/v/billion-context-pi.svg?style=flat-square" alt="npm"></a>
 <a href="https://github.com/ranxianglei/billion-context-pi/blob/master/LICENSE"><img src="https://img.shields.io/npm/l/billion-context-pi.svg?style=flat-square" alt="license"></a>
@@ -76,9 +66,10 @@ pi install npm:billion-context-pi
 
 That's it. The extension auto-loads on next Pi startup. No configuration needed — it reads your model's context window automatically.
 
-> **Already using another sub-agent extension?** Pick one delegation toolset so the model doesn't waffle between two:
-> - **Use ACP's `acp_delegate` instead** (default, recommended) — uninstall the other one, e.g. `pi remove npm:pi-subagents`. It replaces pi-subagents at a fraction of the context cost (~600 tok vs ~7K tok/turn); see below.
-> - **Keep your own sub-agents** — set `"delegate": false` in `.pi/acp.json` (project) or `~/.pi/acp.json` (global): ACP then registers no `acp_delegate*` tools and injects no delegation prompt. See [Using your own sub-agents](#using-your-own-sub-agents).
+> **Uninstall `pi-subagents` first (optional, recommended).** billion-context-pi ships its own `acp_delegate` sub-agent tool (see below) that replaces pi-subagents at a fraction of the context cost (~600 tok vs ~7K tok/turn). If you have pi-subagents installed, remove it to avoid duplicate delegation tools:
+> ```bash
+> pi remove npm:pi-subagents
+> ```
 
 ## How it works
 
@@ -132,8 +123,6 @@ billion-context-pi is built for the **Pi** coding agent (`@earendil-works/pi-cod
 | `acp_delegate_wait` | Block until a delegate run finishes (returns its result; times out otherwise) |
 | `acp_delegate_cancel` | Cancel a running delegate by runId |
 
-The three `acp_delegate*` rows can be removed entirely with `"delegate": false` in `acp.json` — see [Using your own sub-agents](#using-your-own-sub-agents).
-
 ### acp_delegate — clean-context delegation
 
 Hand a self-contained task to a fresh pi process running in a clean context. Five built-in roles, each with a system prompt and a **soft tool guardrail**:
@@ -158,22 +147,6 @@ The full delegate result is saved to a file (`/tmp/acp-delegate/<runId>.out`); t
 
 In the **interactive TUI**, async runs also show a live status widget below the editor (agent, elapsed seconds, task preview), so you always know what's running and for how long. Disabled automatically in RPC/print/JSON.
 
-#### Using your own sub-agents
-
-If you keep your own sub-agent extension (pi-subagents, pi-lens, ...) and don't want ACP's delegation toolset on top of it, disable it:
-
-```json
-{ "delegate": false }
-```
-
-in `.pi/acp.json` (project) or `~/.pi/acp.json` (global); project overrides global. `"delegate": false` is equivalent to `"delegate": { "enabled": false }` — the other keys (`displayUsage`, timeouts, ...) keep their defaults, they just go unused while disabled.
-
-What turns off: the three `acp_delegate*` tools, the `ctrl+alt+f` shortcut, and the `ACP_DELEGATE NOTIFICATIONS` system-prompt section. Everything else (compress / decompress / search_context / acp_status, pruning, nudges) is unaffected.
-
-Timing: tools are registered at session start, so the change takes effect on the **next session** (restart Pi or start a new session); the prompt section is re-resolved every turn and changes immediately within a session. Mid-session edits can therefore briefly desynchronize the two surfaces (enabling shows the prompt before the tools exist; disabling hides the prompt while the tools stay registered until restart).
-
-Related knobs: `delegatePrompt: null` removes only the prompt section and keeps the tools — the inverse of `"delegate": false`, which removes both. Don't use Pi's `--exclude-tools acp_delegate,...` as a substitute for `"delegate": false`: the denylist hides the tools, but the prompt section is driven by config only — the model would still be told about `acp_delegate_wait` etc. even though it cannot call them.
-
 ## `/acp` command
 
 Rich status display for the user:

package/README.zh-CN.md

@@ -8,16 +8,6 @@
 
 ---
 
-## 📄 论文 / 预印本
-
-- **[模型驱动的分层增量压缩:面向长寿命编码 Agent 的免训练多代上下文管理](./paper/模型驱动的分层增量压缩-免训练多代上下文管理.md)**(中文版,v0.2)
-
-> 📝 **论文本身与代码一同以 MIT 许可开源(位于 `paper/` 目录),是代码库的一部分 —— 这是一份活文档,任何人都可以编辑,欢迎提 PR 改进。**
-
-生产规模纵向研究:四个半月、三宿主、174,327 次模型调用、187.6 亿累计输入 token(三宿主合计约 247 亿),204,800-token 窗口零违规,马拉松会话 8,584–12,049 次调用。
-
----
-
 <p align="center">
 <a href="https://www.npmjs.com/package/billion-context-pi"><img src="https://img.shields.io/npm/v/billion-context-pi.svg?style=flat-square" alt="npm"></a>
 <a href="https://github.com/ranxianglei/billion-context-pi/blob/master/LICENSE"><img src="https://img.shields.io/npm/l/billion-context-pi.svg?style=flat-square" alt="license"></a>
@@ -75,9 +65,10 @@ pi install npm:billion-context-pi
 
 完成。扩展在下次 Pi 启动时自动加载。无需配置 —— 它会自动读取模型的上下文窗口。
 
-> **正在使用其他子代理扩展?** 选一套委派工具集,避免模型在两套之间摇摆:
-> - **改用 ACP 的 `acp_delegate`**(默认,推荐)— 卸载另一个,例如 `pi remove npm:pi-subagents`。它以极低的上下文成本(~600 tok vs ~7K tok/轮)替代 pi-subagents,见下文。
-> - **保留你自己的子代理** — 在 `.pi/acp.json`(项目级)或 `~/.pi/acp.json`(全局)中设 `"delegate": false`:ACP 不再注册任何 `acp_delegate*` 工具,也不注入委派提示段。见[使用你自己的子代理](#使用你自己的子代理)。
+> **建议先卸载 `pi-subagents`(可选,推荐)。** billion-context-pi 自带 `acp_delegate` 子代理工具(见下文),以极低的上下文成本(~600 tok vs ~7K tok/轮)替代 pi-subagents。如果你已安装 pi-subagents,卸载它以避免重复的委派工具:
+> ```bash
+> pi remove npm:pi-subagents
+> ```
 
 ## 工作原理
 
@@ -131,8 +122,6 @@ billion-context-pi 面向 **Pi** 编码代理(`@earendil-works/pi-coding-agent`)
 | `acp_delegate_wait` | 阻塞等待委派任务完成(返回结果,否则超时) |
 | `acp_delegate_cancel` | 按 runId 取消正在运行的委派任务 |
 
-三行 `acp_delegate*` 工具可用 `acp.json` 中的 `"delegate": false` 整体移除——见[使用你自己的子代理](#使用你自己的子代理)。
-
 ### acp_delegate — 干净上下文委派
 
 把一个自包含的任务交给一个运行在干净上下文中的新 pi 进程。五个内置角色,各自有系统提示和**软工具护栏**:
@@ -156,22 +145,6 @@ Worker 运行在 Pi 的完整默认工具集上 - 不应用 `--tools` 白名单,
 
 在**交互 TUI** 中,异步运行还会在编辑器下方显示一个实时状态 widget(角色、已运行秒数、任务预览),让你随时知道什么在跑、跑了多久。RPC/print/JSON 模式自动禁用。
 
-#### 使用你自己的子代理
-
-如果你保留自己的子代理扩展(pi-subagents、pi-lens 等),不想再叠加 ACP 的委派工具集,关掉它:
-
-```json
-{ "delegate": false }
-```
-
-写入 `.pi/acp.json`(项目级)或 `~/.pi/acp.json`(全局);项目级覆盖全局。`"delegate": false` 与 `"delegate": { "enabled": false }` 等价——其余键(`displayUsage`、超时等)保持默认值,只是禁用期间不生效。
-
-关闭的具体面:`acp_delegate*` 三个工具、`ctrl+alt+f` 快捷键、系统提示中的 `ACP_DELEGATE NOTIFICATIONS` 段。其余功能(compress / decompress / search_context / acp_status、prune、nudge)不受影响。
-
-生效时机:工具在会话开始时注册,改动在**下一个会话**生效(重启 Pi 或开新会话);提示段每回合重新解析,会在当前会话内立即变化。因此会话中途改配置时两个面可能短暂不同步(开启后提示先于工具出现;关闭后提示先消失而工具保留到重启)。
-
-相关开关:`delegatePrompt: null` 只摘除提示段、保留工具——与 `"delegate": false` 全关正好相反。不要用 Pi 的 `--exclude-tools acp_delegate,...` 替代 `"delegate": false`:denylist 只让工具消失,提示段只由配置驱动——模型仍会被告知 `acp_delegate_wait` 等它调不到的工具。
-
 ## `/acp` 命令
 
 为用户提供丰富的状态显示:

package/dist/compress-tool.d.ts

@@ -1,7 +1,6 @@
 import { Type, type Static } from "typebox";
 import type { ToolDefinition } from "@earendil-works/pi-coding-agent";
 import type { AcpRuntime } from "./runtime.js";
-import { type ToolPromptOverrides } from "./surface.js";
 import { type CompressionBlock, type CompressionState } from "acp-kernel";
 declare const RangeSpec: Type.TObject<{
     startId: Type.TString;
@@ -20,7 +19,7 @@ declare const CompressParams: Type.TObject<{
     summaryMaxChars: Type.TOptional<Type.TNumber>;
 }>;
 type CompressArgs = Static<typeof CompressParams>;
-export declare function makeCompressTool(runtime: AcpRuntime, overrides?: ToolPromptOverrides): ToolDefinition<typeof CompressParams>;
+export declare function makeCompressTool(runtime: AcpRuntime): ToolDefinition<typeof CompressParams>;
 type RangeEntry = Static<typeof RangeSpec>;
 export declare function normalizeRanges(args: CompressArgs): RangeEntry[] | string;
 export declare function tailRepair(s: string): string | undefined;

package/dist/config.d.ts

@@ -2,8 +2,6 @@ import { type Config, type Prompts } from "acp-kernel";
 import type { CompressReasoningConfig } from "./reasoning-drop.js";
 import type { DegenerationGuardConfig } from "./degeneration.js";
 import type { ThrottleRetryConfig } from "./throttle-retry.js";
-import type { PiPromptSections } from "./system-prompt.js";
-import type { NudgeSectionsConfig, ToolPromptsConfig } from "./surface.js";
 /** Per-role delegate defaults. Lets long-lived automation pin a cheaper or
  *  more capable model and a thinking level per delegate role, so the main
  *  agent doesn't have to fill them in on every `acp_delegate()` call. */
@@ -122,13 +120,6 @@ export interface CompressSettings {
      *  tool calls — see CompressReasoningConfig in src/reasoning-drop.ts.
      *  Merged field-wise (drop, threshold) across the three levels. */
     reasoning?: CompressReasoningConfig;
-    /** Active prompt pack name (see CONFIGURATION.md “Prompt packs”). Base
-     *  level; override per provider/model via `providers`. "default" or unset =
-     *  built-in defaults. Resolved per request against the live model, so
-     *  switching models mid-session switches the pack. Packs ship text-level
-     *  overrides only; a pack's `toolPrompts` follow the base selection (tool
-     *  definitions freeze at extension load, before the model is known). */
-    promptPack?: string;
 }
 /** Per-provider compression overrides. Carries the same tuning fields as the
  *  global level, plus an optional per-model map keyed by model id. */
@@ -261,21 +252,6 @@ export interface AdapterConfig {
      *  replacing the kernel's tuned compression rules may reduce summary quality
      *  (lost paths/signatures/decisions → worse retrieval). */
     acknowledgePromptsRisk?: boolean;
-    /** Override structural sections of the ACP system prompt (ACP TAGS, TOOLS,
-     *  WHEN TO COMPRESS, ...). Tri-state per section: string = replace, null =
-     *  remove, omitted = default. Not risk-gated — these are documentation
-     *  sections, not compression rules. Set via acp.json. */
-    promptSections?: Partial<PiPromptSections>;
-    /** Override guidance-class nudge texts (efficiencyNote, emergencyHeader,
-     *  t2Guidance, t3Guidance). Same tri-state semantics. Not risk-gated. */
-    nudgeSections?: NudgeSectionsConfig;
-    /** Override the four ACP tool definitions' LLM-facing text (description,
-     *  paramDescriptions, promptSnippet, promptGuidelines). Read synchronously
-     *  at extension load — tool defs are frozen at registration time. */
-    toolPrompts?: ToolPromptsConfig;
-    /** Replace (string) or remove (null) the ACP_DELEGATE_NOTIFICATIONS appendix
-     *  injected when the delegate tool is enabled. */
-    delegatePrompt?: string | null;
     coreOverrides?: Partial<Config>;
 }
 export declare const DEFAULT_TOOL_BASH_TIMEOUT = 60;

package/dist/decompress-tool.d.ts

@@ -1,12 +1,11 @@
 import { Type } from "typebox";
 import type { ToolDefinition } from "@earendil-works/pi-coding-agent";
 import type { AcpRuntime } from "./runtime.js";
-import { type ToolPromptOverrides } from "./surface.js";
 declare const DecompressParams: Type.TObject<{
     blockId: Type.TString;
     full: Type.TOptional<Type.TBoolean>;
     toFile: Type.TOptional<Type.TString>;
     inline: Type.TOptional<Type.TBoolean>;
 }>;
-export declare function makeDecompressTool(runtime: AcpRuntime, overrides?: ToolPromptOverrides): ToolDefinition<typeof DecompressParams>;
+export declare function makeDecompressTool(runtime: AcpRuntime): ToolDefinition<typeof DecompressParams>;
 export {};