docs: reserve seam for complete capabilities
This commit is contained in:
@@ -2,5 +2,5 @@
|
||||
# side as of the last confirmed-consistent state. Both languages carry equal authority;
|
||||
# after editing either side, bring the other along and re-record with:
|
||||
# pnpm run verify-translation-pairing --write packages/compact/README.md
|
||||
README.md: 0a83b87f57e4c948ee6df9e622f17964aff00f0f
|
||||
README.zh.md: 001ca6f7c079a60cbd432664e4f0b06955aaf487
|
||||
README.md: 0b5a5b18b963dc7be0bd314420bdb3c88715533d
|
||||
README.zh.md: 5d6d8a1c076e444094cc030c9495f346f40bf7ea
|
||||
|
||||
@@ -2,7 +2,7 @@
|
||||
|
||||
English | [中文](README.zh.md)
|
||||
|
||||
A compaction capability family (see [capability seams](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)): an abstract interface, a summarizing backend, a model-free tool-result pruning companion, and a human command adapter. All **product** packages.
|
||||
A compaction capability family (see [capability seams](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)): a Service Definition, a summarizing provider, a model-free tool-result pruning companion, and a human command Consumer. All **product** packages.
|
||||
|
||||
| Package | Role | ctx key |
|
||||
|---|---|---|
|
||||
|
||||
@@ -2,7 +2,7 @@
|
||||
|
||||
[English](README.md) | 中文
|
||||
|
||||
一个压缩(compaction)能力家族(参见[能力 seam](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)):抽象接口、摘要后端、无模型工具结果修剪配套工具,以及用户命令适配器。这些全是**产品**包。
|
||||
一个压缩(compaction)能力家族(参见[能力 seam](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)):Service Definition、摘要提供方、无模型工具结果修剪配套工具,以及人类命令 Consumer。这些全是**产品**包。
|
||||
|
||||
| 包 | 职责 | ctx key |
|
||||
|---|---|---|
|
||||
|
||||
@@ -2,5 +2,5 @@
|
||||
# side as of the last confirmed-consistent state. Both languages carry equal authority;
|
||||
# after editing either side, bring the other along and re-record with:
|
||||
# pnpm run verify-translation-pairing --write packages/compact/compact-basic/README.md
|
||||
README.md: 5395219147e8932c8e5874e84e534a2c2fe33d53
|
||||
README.zh.md: 59479b26c5c8cf27e6ac84814c09f14ecbf9662c
|
||||
README.md: e0e3413d42e6afb8e95f91d79f9bdf3a1f17fbda
|
||||
README.zh.md: 4515592710f855dc713f078479ebe7c9c7299de1
|
||||
|
||||
@@ -2,9 +2,9 @@
|
||||
|
||||
English | [中文](README.zh.md)
|
||||
|
||||
The **basic compaction backend**: a `BasicCompactService` implementing the `@deepseek-ai/dsh-compact` seam with reusable `ctx.tokenMeter` pressure, token-budget retention, and summarization as a direct one-shot `ctx.llm.stream()` call that replays the conversation prefix to reuse the provider's KV cache (interceptable at `llm/stream`).
|
||||
The **basic compaction backend**: a `BasicCompactService` implementing the `@deepseek-ai/dsh-compact` Service Definition with reusable `ctx.tokenMeter` pressure, token-budget retention, and summarization as a direct one-shot `ctx.llm.stream()` call that replays the conversation prefix to reuse the provider's KV cache (interceptable at `llm/stream`).
|
||||
|
||||
This is the implementation tier of the compaction capability — see the [interface package](../compact/README.md) for the seam and the [capability-seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md) for the design.
|
||||
This package owns the Service provider role of the compaction capability — see the [Service Definition package](../compact/README.md) for its contract and the [capability-seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md) for the design.
|
||||
|
||||
## What it owns
|
||||
|
||||
@@ -15,7 +15,7 @@ This backend owns the compaction policy:
|
||||
- **Model-free pruning** — after pressure or canonical overflow qualifies, the optional [`ctx.toolResultPrune`](../compact-tool-result-prune/README.md) service rewrites oversized tool results before range selection. Compact-basic remeasures through `ctx.tokenMeter`, skips summarization when pressure becomes safe, and otherwise summarizes the pruned surface. Below-pressure step checks never prune.
|
||||
- **Retention** — compact the oldest whole surface units while preserving a recent tail and balanced tool-call/result cuts through the [`dsh-compact` boundary helpers](../compact/README.md#tool-pairing-boundaries). Turn boundaries do not protect old steps inside a runaway turn. An open indivisible tail declines until it closes. The optional pruner can repair an oversized closed tool unit when its text-bearing result is the removable bulk; indivisible non-tool units and non-prunable tool remainders remain out of scope.
|
||||
- **Convergence** — retry head-checkpoint compaction up to `compactionRetries`; reject a summary that does not shrink its source, and throw if retries cannot return below threshold.
|
||||
- **Summarization** — a direct `llm/stream` call uses the configured provider/model pair and cap, falling back to the latest logged request target and then the agent target, without running the loop-only `agent/request` seam. The call replays the conversation's own system prompt, tools, and shadowed-region messages verbatim and appends the compaction instruction as the final user message, so it reuses the provider's warm prefix cache instead of invalidating it. It sets `GenerateOptions.purpose` to `compaction`, which adapters may forward as request attribution (the DeepSeek adapter sends `x-deepseek-harness-compact: 1`) without touching the model-visible body. Only returned text enters the checkpoint, excluding reasoning and tool calls that would leak private reasoning or create an orphaned call.
|
||||
- **Summarization** — a direct `llm/stream` call uses the configured provider/model pair and cap, falling back to the latest logged request target and then the agent target, without running the loop-only `agent/request` extension point. The call replays the conversation's own system prompt, tools, and shadowed-region messages verbatim and appends the compaction instruction as the final user message, so it reuses the provider's warm prefix cache instead of invalidating it. It sets `GenerateOptions.purpose` to `compaction`, which adapters may forward as request attribution (the DeepSeek adapter sends `x-deepseek-harness-compact: 1`) without touching the model-visible body. Only returned text enters the checkpoint, excluding reasoning and tool calls that would leak private reasoning or create an orphaned call.
|
||||
- **Framing** — the replacement user message marks established checkpoint context with `<compacted-summary>` tags. The raw summary remains on the `compact/summary` event, and later automatic cycles merge the prior checkpoint.
|
||||
- **Lifecycle** — all entry points share one bracket-first region transaction. It validates the range and live lock, appends `compact/start` synchronously, prepares and awaits the summary, revalidates, appends `compact/summary` plus the replacement, and makes exactly one closing attempt. Automatic and explicit-region calls require a numeric open-turn owner and whole-surface stability; the serial `agent/pre-step` listener checks pressure before request derivation, while canonical provider overflow enters through `agent/request-error` and authorizes retry only after durable surface progress. `compactNow()` reserves idle admission, uses `turn: null`, accepts append-only context outside its selected span, flushes every closed attempt, and releases admission in `finally`.
|
||||
- **Overflow recovery** — provider-confirmed overflow needs no capacity metadata: it bypasses normal pressure and retention, prunes, then attempts one maximal balanced head reduction while leaving the newest indivisible unit. Retry is authorized whenever `surface.replaceGeneration` advances, including when pruning lands before later summary work throws. No replacement, an exhausted target-specific cap, cancellation, or an unknown/noncanonical error preserves the original provider failure.
|
||||
|
||||
@@ -2,9 +2,9 @@
|
||||
|
||||
[English](README.md) | 中文
|
||||
|
||||
**基础压缩(compaction)后端**:`BasicCompactService` 实现 `@deepseek-ai/dsh-compact` seam,使用可复用的 `ctx.tokenMeter` 压力、token 预算保留与摘要。摘要是直接的一次性 `ctx.llm.stream()` 调用,它会回放会话前缀以复用提供方的 KV Cache(可在 `llm/stream` 处拦截)。
|
||||
**基础压缩(compaction)后端**:`BasicCompactService` 实现 `@deepseek-ai/dsh-compact` Service Definition,使用可复用的 `ctx.tokenMeter` 压力、token 预算保留与摘要。摘要是直接的一次性 `ctx.llm.stream()` 调用,它会回放会话前缀以复用提供方的 KV Cache(可在 `llm/stream` 处拦截)。
|
||||
|
||||
这是压缩能力的实现层。seam 见 [接口包](../compact/README.md),设计见 [能力 seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md)。
|
||||
本包承担压缩能力的 Service provider 角色;其约定见 [Service Definition 包](../compact/README.md),设计见 [能力 seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md)。
|
||||
|
||||
## 拥有的职责
|
||||
|
||||
@@ -15,7 +15,7 @@
|
||||
- **不依赖模型的剪枝**:在压力或规范溢出符合条件后,可选的 [`ctx.toolResultPrune`](../compact-tool-result-prune/README.md) 服务会在选择范围之前改写超大工具结果。Compact-basic 通过 `ctx.tokenMeter` 重新测量;如果压力已回到安全范围,就跳过摘要,否则对已剪枝的表层进行摘要。低于压力的步骤检查绝不剪枝。
|
||||
- **保留**:压缩最旧的完整表层单元,同时保留近期尾部,并通过 [`dsh-compact` 边界 helper](../compact/README.md#tool-pairing-boundaries) 将切分点调整到工具调用/结果配对平衡的位置。轮次边界不会保护失控轮次内的旧步骤。尚未闭合且不可分的尾部会在闭合前拒绝压缩。当闭合的超大工具单元以文本型结果为可移除主体时,可选 pruner 可以修复它;不可分的非工具单元与不可剪枝的工具剩余部分不在范围内。
|
||||
- **收敛**:最多按 `compactionRetries` 重试头部检查点压缩;拒绝不能缩小源内容的摘要,如果重试仍无法回到阈值以下,则抛出异常。
|
||||
- **摘要**:直接 `llm/stream` 调用使用已配置的提供方/模型对与上限,回退到最新已记录请求目标,然后再回退到 agent 目标,而不运行仅用于 agent loop 的 `agent/request` seam。该调用会逐字回放会话自身的系统提示词、工具与已遮蔽区域消息,并将压缩指令作为最后一条 user 消息追加,从而复用提供方的热前缀 cache,而非使它失效。它将 `GenerateOptions.purpose` 设为 `compaction`,适配器可将其作为请求归因转发(DeepSeek 适配器发送 `x-deepseek-harness-compact: 1`),但不会触碰模型可见的请求体。只有返回的文本会进入检查点;推理(reasoning)和工具调用都会被排除,以免泄露私有推理或产生遗留调用。
|
||||
- **摘要**:直接 `llm/stream` 调用使用已配置的提供方/模型对与上限,回退到最新已记录请求目标,然后再回退到 agent 目标,而不运行仅用于 agent loop 的 `agent/request` 扩展点。该调用会逐字回放会话自身的系统提示词、工具与已遮蔽区域消息,并将压缩指令作为最后一条 user 消息追加,从而复用提供方的热前缀 cache,而非使它失效。它将 `GenerateOptions.purpose` 设为 `compaction`,适配器可将其作为请求归因转发(DeepSeek 适配器发送 `x-deepseek-harness-compact: 1`),但不会触碰模型可见的请求体。只有返回的文本会进入检查点;推理(reasoning)和工具调用都会被排除,以免泄露私有推理或产生遗留调用。
|
||||
- **框定**:替换 user 消息使用 `<compacted-summary>` 标签标记已建立的检查点上下文。原始摘要保留在 `compact/summary` 事件上,后续自动周期会合并之前的检查点。
|
||||
- **生命周期**:所有入口点共享一个先记录标记的区域事务。它会验证范围与活动锁,同步追加 `compact/start`,准备并等待摘要,重新验证,再追加 `compact/summary` 和替换,最后恰好进行一次闭合尝试。自动调用和显式范围调用要求数字标识的开放轮次归属,并要求整个表层保持稳定;串行 `agent/pre-step` listener 会在派生请求之前检查压力,而规范提供方溢出则经由 `agent/request-error` 进入,并且只在表层取得持久进展后才允许重试。`compactNow()` 会预留空闲接纳,使用 `turn: null`,允许所选 span 之外追加仅追加上下文,flush 每次已闭合尝试,并在 `finally` 中释放接纳预留。
|
||||
- **溢出恢复**:提供方已确认的溢出不需容量元数据。它会绕过常规压力与保留,执行剪枝,再尝试一次最大平衡头部缩减,并留下最新不可分单元。只要 `surface.replaceGeneration` 前进,就允许重试,包括剪枝在后续摘要工作抛出异常前已落地的情况。如果没有替换、目标特定上限已耗尽、已取消,或遇到未知/非规范错误,则保留原始提供方失败。
|
||||
|
||||
@@ -2,5 +2,5 @@
|
||||
# side as of the last confirmed-consistent state. Both languages carry equal authority;
|
||||
# after editing either side, bring the other along and re-record with:
|
||||
# pnpm run verify-translation-pairing --write packages/compact/compact/README.md
|
||||
README.md: c7fdbef33b2e65089a0f0112e19122db4f26bb77
|
||||
README.zh.md: 493d269202b3a5b89cd600391e7144219ed2bf12
|
||||
README.md: 3fecc91422bcdc994bcbe4db308ee4a52cb3c53f
|
||||
README.zh.md: ea66d8bdb0684f27cf68a5431be75966041b7534
|
||||
|
||||
@@ -4,19 +4,19 @@ English | [中文](README.zh.md)
|
||||
|
||||
The **`CompactService`** (`ctx.compact`) defines WHAT compaction does — decide when history is too large and summarize an older range into a single surface node — without saying HOW.
|
||||
|
||||
This package is the interface tier of the compaction capability, split so each concern evolves (and swaps) independently:
|
||||
This package owns the Service Definition role of the compaction capability, split so each role evolves (and swaps) independently:
|
||||
|
||||
| Package | Role |
|
||||
|---|---|
|
||||
| `@deepseek-ai/dsh-compact` (this) | the interface: abstract service + `compact/*` events + `CompactionResult` + canonical checkpoint source + tool-pairing boundary helpers |
|
||||
| `@deepseek-ai/dsh-compact-basic` | a backend: `ctx.tokenMeter` pressure + token-budget retention + `llm.stream()` summarization |
|
||||
| `@deepseek-ai/dsh-command-compact` | the human `/compact` command over `ctx.compact.compactNow()` |
|
||||
| `@deepseek-ai/dsh-compact` (this) | Service Definition: abstract service + `compact/*` events + `CompactionResult` + canonical checkpoint source + tool-pairing boundary helpers |
|
||||
| `@deepseek-ai/dsh-compact-basic` | Service provider: `ctx.tokenMeter` pressure + token-budget retention + `llm.stream()` summarization |
|
||||
| `@deepseek-ai/dsh-command-compact` | Consumer: the human `/compact` command over `ctx.compact.compactNow()` |
|
||||
|
||||
Unlike the bash seam, this interface depends on `@deepseek-ai/dsh-session` and `@deepseek-ai/dsh-llm` — the contract's verbs are defined over a `Session` and its output is the `ContentBlock` vocabulary, so they cannot be expressed without naming those packages. That deviation from the "interface depends only on cordis" guidance is intentional and recorded in the [compaction capability-seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md).
|
||||
Unlike the bash seam, this Service Definition depends on `@deepseek-ai/dsh-session` and `@deepseek-ai/dsh-llm` — the contract's verbs are defined over a `Session` and its output is the `ContentBlock` vocabulary, so they cannot be expressed without naming those packages. That deviation from the "Service Definition depends only on cordis" guidance is intentional and recorded in the [compaction capability-seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md).
|
||||
|
||||
## Service API (`ctx.compact`)
|
||||
|
||||
All three operations are **abstract** — the backend owns trigger policy, retention, event sequencing, and summarization. Reusable request measurement is a separate service, [`ctx.tokenMeter`](../../llm/token-meter/README.md), rather than part of this interface.
|
||||
All three operations are **abstract** — the backend owns trigger policy, retention, event sequencing, and summarization. Reusable request measurement is a separate service, [`ctx.tokenMeter`](../../llm/token-meter/README.md), rather than part of this Service Definition.
|
||||
|
||||
| Member | Semantics |
|
||||
|---|---|
|
||||
@@ -32,7 +32,7 @@ All three operations are **abstract** — the backend owns trigger policy, reten
|
||||
|
||||
## Tool-pairing boundaries
|
||||
|
||||
The interface exports `toolPairingBalancedBefore(session, seq)` and `toolPairingBalancedAfter(session, seq)` for snapping and validating compaction edges. A safe edge has no unanswered assistant tool call crossing it. Each helper validates that the event sequence is in the current surface and answers from balances cached per cut in surface order.
|
||||
The Service Definition exports `toolPairingBalancedBefore(session, seq)` and `toolPairingBalancedAfter(session, seq)` for snapping and validating compaction edges. A safe edge has no unanswered assistant tool call crossing it. Each helper validates that the event sequence is in the current surface and answers from balances cached per cut in surface order.
|
||||
|
||||
The private per-session cache is keyed by `session.surface.replaceGeneration` and the processed surface-entry count. An unchanged generation extends the fold with unseen tail entries only; a log-only append with no new surface entry does no event reads, while a replacement generation rebuilds current membership and balances. Missing event seqs and a `tool/result` without a preceding open call reject as corrupt surface state.
|
||||
|
||||
@@ -80,7 +80,7 @@ A successful implementation replaces an older surface range with one user-role s
|
||||
|
||||
#### Token effect
|
||||
|
||||
Zero direct tokens from this interface. A backend trades many retained history tokens for one summary and leaves the recent tail unchanged.
|
||||
Zero direct tokens from this Service Definition. A backend trades many retained history tokens for one summary and leaves the recent tail unchanged.
|
||||
|
||||
#### KV Cache effect
|
||||
|
||||
|
||||
@@ -4,19 +4,19 @@
|
||||
|
||||
**`CompactService`**(`ctx.compact`)定义压缩做什么,即判定历史记录是否过大,并将较早范围摘要为单个表层节点,但不规定如何实现。
|
||||
|
||||
这个包是压缩能力的接口层,因此各项职责均可独立演进,也可独立替换:
|
||||
本包承担压缩能力的 Service Definition 角色,因此各角色均可独立演进,也可独立替换:
|
||||
|
||||
| 包 | 职责 |
|
||||
|---|---|
|
||||
| `@deepseek-ai/dsh-compact`(本包) | 接口:抽象服务 + `compact/*` 事件 + `CompactionResult` + 规范检查点源 + 工具配对边界 helper |
|
||||
| `@deepseek-ai/dsh-compact-basic` | 后端:`ctx.tokenMeter` 压力 + token 预算保留 + `llm.stream()` 摘要 |
|
||||
| `@deepseek-ai/dsh-command-compact` | 面向用户的 `/compact` 命令,基于 `ctx.compact.compactNow()` 实现 |
|
||||
| `@deepseek-ai/dsh-compact`(本包) | Service Definition:抽象服务 + `compact/*` 事件 + `CompactionResult` + 规范检查点源 + 工具配对边界 helper |
|
||||
| `@deepseek-ai/dsh-compact-basic` | Service provider:`ctx.tokenMeter` 压力 + token 预算保留 + `llm.stream()` 摘要 |
|
||||
| `@deepseek-ai/dsh-command-compact` | Consumer:面向人类的 `/compact` 命令,基于 `ctx.compact.compactNow()` 实现 |
|
||||
|
||||
与 bash seam 不同,该接口依赖 `@deepseek-ai/dsh-session` 和 `@deepseek-ai/dsh-llm`。约定的动词基于 `Session` 定义,其输出使用 `ContentBlock` 词汇,因此无法在不指名这些包的情况下表达。这项对「接口只依赖 cordis」指引的偏离是有意的,并记录在 [压缩能力 seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md) 中。
|
||||
与 bash seam 不同,该 Service Definition 依赖 `@deepseek-ai/dsh-session` 和 `@deepseek-ai/dsh-llm`。约定的动词基于 `Session` 定义,其输出使用 `ContentBlock` 词汇,因此无法在不指名这些包的情况下表达。这项对「Service Definition 只依赖 cordis」指引的偏离是有意的,并记录在 [压缩能力 seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md) 中。
|
||||
|
||||
## 服务 API(`ctx.compact`)
|
||||
|
||||
三个操作都是**抽象方法**:触发策略、保留、事件顺序与摘要均属于后端。可复用的请求测量是独立服务 [`ctx.tokenMeter`](../../llm/token-meter/README.md),而非本接口的一部分。
|
||||
三个操作都是**抽象方法**:触发策略、保留、事件顺序与摘要均属于后端。可复用的请求测量是独立服务 [`ctx.tokenMeter`](../../llm/token-meter/README.md),而非本 Service Definition 的一部分。
|
||||
|
||||
| 成员 | 语义 |
|
||||
|---|---|
|
||||
@@ -32,7 +32,7 @@
|
||||
|
||||
## 工具配对边界
|
||||
|
||||
该接口导出 `toolPairingBalancedBefore(session, seq)` 与 `toolPairingBalancedAfter(session, seq)`,用于对齐和验证压缩边界。安全边界不会被尚未回答的 assistant 工具调用跨越。每个 helper 都会验证给定事件 seq 位于当前表层,并根据按表层顺序缓存的各切分点配对状态返回结果。
|
||||
该 Service Definition 导出 `toolPairingBalancedBefore(session, seq)` 与 `toolPairingBalancedAfter(session, seq)`,用于对齐和验证压缩边界。安全边界不会被尚未回答的 assistant 工具调用跨越。每个 helper 都会验证给定事件 seq 位于当前表层,并根据按表层顺序缓存的各切分点配对状态返回结果。
|
||||
|
||||
每个会话的私有 cache 以 `session.surface.replaceGeneration` 和已处理表层条目数为 key。generation 未变时,只需将尚未处理的尾部条目纳入累计结果;仅向日志追加、但未新增表层条目时,不会读取事件。replace generation 变化时则会重建当前成员关系与配对状态。事件 seq 缺失以及 `tool/result` 没有对应的先前未闭合调用,均会被视为表层状态损坏并遭拒绝。
|
||||
|
||||
@@ -80,7 +80,7 @@
|
||||
|
||||
#### Token 影响
|
||||
|
||||
该接口不会直接产生 token。后端用一份摘要换取多个原本保留的历史 token,并保持近期尾部不变。
|
||||
该 Service Definition 不会直接产生 token。后端用一份摘要换取多个原本保留的历史 token,并保持近期尾部不变。
|
||||
|
||||
#### KV Cache 影响
|
||||
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/**
|
||||
* Compaction service seam (`ctx.compact`): implementations decide when to
|
||||
* Compaction Service Definition (`ctx.compact`): providers decide when to
|
||||
* compact and replace a history range with one summary node by subclassing
|
||||
* {@link CompactService}. This interface necessarily depends on session and LLM
|
||||
* vocabulary; the rationale is in the
|
||||
|
||||
@@ -13,7 +13,7 @@ import type { ManualCompactAgentContext } from '@deepseek-ai/dsh-compact'
|
||||
|
||||
/**
|
||||
* A trivial concrete CompactService implementing the abstract contract. The
|
||||
* interface package owns no algorithm — these tests exercise the seam itself:
|
||||
* Service Definition package owns no algorithm — these tests exercise its contract:
|
||||
* service registration, the abstract method shape, and the `compact/*` event
|
||||
* declaration merge.
|
||||
*/
|
||||
|
||||
Reference in New Issue
Block a user