feat(system-prompt): cache dynamic policy context

.agents/notes/implemented/feature/2026-07-30-current-sandbox-policy-context.i18n.yaml

@@ -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 .agents/notes/implemented/feature/2026-07-30-current-sandbox-policy-context.md
-2026-07-30-current-sandbox-policy-context.md: 380c00962c54ba06aed2fe452673472a7dfdc7c4
-2026-07-30-current-sandbox-policy-context.zh.md: 2e76fd9e3193014b99d45dfea662eafd3008343a
+2026-07-30-current-sandbox-policy-context.md: 16805cc16242dca825dcc58b1e0c7baa4b1fbd0d
+2026-07-30-current-sandbox-policy-context.zh.md: 9ef0dcc268ecbd78b2fb384ac09fd39f6b6c86d5

.agents/notes/implemented/feature/2026-07-30-current-sandbox-policy-context.md

@@ -10,15 +10,19 @@ The sandbox policy already enforced and logged each session's file-effect mode,
 
 ## Decision
 
-`dsh-sandbox-policy`, the owner of mode and workspace-root resolution, registers one `sandbox:policy` system-prompt section. Every agent request resolves the active session directly through `ctx.sandboxPolicy.resolve({ session })`; there is no denial-history scan, delta narrator, or in-memory “last told” state.
+`dsh-sandbox-policy`, the owner of mode and workspace-root resolution, registers one `sandbox:policy` cache-safe context contribution. Every agent request resolves the active session directly through `ctx.sandboxPolicy.resolve({ session })`; there is no denial-history scan or process-local “last told” state.
 
-Enforcing backends register independently disposable `filesystem`, `bash`, or `terminal` family contributions with the policy owner. The section names only registered families in canonical order, and is empty without one. This is current need, not a future extension: the shipped headless inheritance composition combines sandboxed filesystem tools with unfenced one-shot bash, while the persistent-tools composition combines sandboxed filesystem tools and terminal commands without a sandboxed one-shot bash executor. A blanket statement would be false in both.
+Enforcing backends register independently disposable `filesystem`, `bash`, or `terminal` family contributions with the policy owner. The contribution names only registered families in canonical order, and is empty without one. This is current need, not a future extension: the shipped headless inheritance composition combines sandboxed filesystem tools with unfenced one-shot bash, while the persistent-tools composition combines sandboxed filesystem tools and terminal commands without a sandboxed one-shot bash executor. A blanket statement would be false in both.
 
-The section states only facts shared by every enforcement dialect for each registered family. `read-only` says those operations cannot modify files. `workspace-write` states the canonical session workspace with non-exclusive wording and summarizes, without enumerating, that some platform temporary areas may also be writable. `danger-full-access` says the DSH file sandbox does not restrict those operations. Backend-selected temporary paths, `/dev/null`, runner readiness, and other policy domains are absent because `resolve()` cannot establish them at request assembly.
+The contribution states only facts shared by every enforcement dialect for each registered family. `read-only` says those operations cannot modify files in the standing mode. Tool owners separately register a family only when their real schema and execution path offer an approved wider retry; the read-only text adds anti-refusal guidance only for the intersection of enforced and escalatable families. `workspace-write` states the canonical session workspace with non-exclusive wording and summarizes, without enumerating, that some platform temporary areas may also be writable. `danger-full-access` says the DSH file sandbox does not restrict those operations. Backend-selected temporary paths, `/dev/null`, runner readiness, and other policy domains are absent because `resolve()` cannot establish them at request assembly.
 
-The provider runs during normal request assembly, after a `/permission` switch has committed its existing `sandbox/mode` event and before `request/header` is logged. The rendered system text is therefore the durable reconstruction of the exact model-visible fact. Repeated assemblies over unchanged session state produce identical bytes; resume and replay fold the same durable mode event and immutable `SessionHeader.cwd` without catch-up state.
+The existing `dsh-system-prompt` assembly now has ordered dynamic contexts alongside stable system sections and tool schemas. After assembling one step, agent-loop renders all active contexts as one full snapshot with an explicit supersession statement. It appends a sourced `user/message` only when no retained snapshot exists, the bytes changed, compaction removed the retained message, or the final contribution disappeared and needs one clearing snapshot. The snapshot is appended after existing history and before `step/start`, so a changed policy preserves the preceding system-and-conversation cache prefix. The session event itself reconstructs the exact model input; `request/header` remains byte-identical when only policy context changes.
 
-Ownership stays narrow. Approval policy remains the separate `approval:policy` section, plan mode remains `plan:policy`, and tool plugins continue to own schemas plus attempt, denial, and escalation guidance. The prompt states standing policy; filesystem, one-shot bash, and terminal backends remain the enforcement boundaries.
+Ownership stays narrow. Approval policy contributes its complete current `ask` or `never` fact to the same full snapshot; migrating sandbox alone would not preserve cache because `/permission` changes both owners. Plan mode remains `plan:policy`, and tool plugins continue to own schemas plus attempt, denial, and escalation guidance. Context states standing policy; filesystem, one-shot bash, and terminal backends remain the enforcement boundaries.
+
+The cache decision follows current source rather than analogy alone. Codex models permissions as a developer-role `WorldState` section with a persisted fingerprint, emits it only when state changes or retained history lost the fragment, and records the snapshot transition. Hermes keeps its system prompt fixed for a session and explicitly prepends changing skill, model, and voice notices to the next user message to avoid invalidating prompt cache. Pi has no comparable built-in sandbox state, and Claude Code's current native implementation is not publicly inspectable; Anthropic's public cache guidance nevertheless places changing per-request context after the stable cached prefix.
+
+The earlier real-provider Web fixture quantified the defect in the system-section version. The first `danger-full-access` and `workspace-write` requests each reported only 256 cache-read tokens against 14,691 and 14,782 uncached input tokens. Later steps under an unchanged policy reported approximately 14.7k–15.5k cache-read tokens. Moving only the sandbox sentence would not fix those misses because the same preset switch also rewrote the approval-policy system section.
 
 ## Wording evidence
 
@@ -26,24 +30,32 @@ The wording experiment pre-registered preemptive refusal as its primary endpoint
 
 The positive control therefore failed the pre-registered sensitivity gate. Candidate A and B were not run, and this experiment does not select or validate the current wording. It instead establishes that the earlier five-of-twelve result is not reproducible under this task and current tool guidance, and that a stronger positive control or different task distribution is required before making model-behavior rate claims. Deterministic tests below establish truthful request construction and replay only.
 
+The cache-safe delivery rework then supplied a separate, non-statistical acceptance comparison over the neutral Web task `Create the relative path policy-neutral.txt ...`; it does not replace the pre-registered twelve-session experiment. Candidate A's categorical read-only statement produced a text refusal with zero tool calls. Candidate B added one composition-conditioned sentence only for enforced families whose tools expose escalation. A fresh real-provider run then issued an ordinary `write`, observed the read-only denial, retried the same operation in the same turn with `sandbox_permissions: "workspace-write"`, received approval, read the file back, and verified the exact contents. It made no speculative escalation. Across the permission switches and four mutation steps, cache reads were 14,848–15,872 tokens while uncached input was 59–251 tokens per request, directly demonstrating the stable-prefix benefit.
+
 ## Alternatives considered
 
 **Narrate only mode changes.** Rejected because it leaves a fresh session uninformed and makes the first denied operation the policy-discovery mechanism. It also requires a baseline definition that is unnecessary when current state can be rendered directly.
 
 **Scan denial history or remember the last narrated mode.** Rejected because denial events describe attempted operations, not authoritative current state, while process-local bookkeeping does not survive resume. The owner can fold the durable policy directly on every request.
 
-**A generic runtime-facts registry.** Rejected because the existing system-prompt registry already evaluates owner-provided sections with the live agent at request time. One policy owner has no cross-domain invariant that justifies another package or registry.
+**Put current policy in a dynamic system section.** Rejected after real provider evidence showed that a first-time permission switch reduced cache reads to 256 tokens while roughly 14.7k input tokens missed. DeepSeek matches complete prefixes; changing the first wire message prevents reuse of the longer system-plus-history prefix.
 
-**Repeat tool schemas or approval and plan guidance in the section.** Rejected because those surfaces already have owners and independent lifecycles. Duplicating them would create contradictory request prefixes and broaden invalidation.
+**Call `agent.inject()` independently from each policy owner.** Rejected because sibling listener order would define model order, separate messages could expose mismatched intermediate snapshots, and every owner would need its own compaction-retention scan. The existing assembly owner can order contributions and materialize one atomic full snapshot.
+
+**A generic runtime-facts package.** Rejected because the existing system-prompt assembly already owns sections, schemas, variables, scope, and the authoritative per-step waterfall. Extending that owner with ordered contexts adds no package or second registry service.
+
+**Repeat tool schemas or plan guidance in the context.** Rejected because those surfaces already have owners and independent lifecycles. Approval current state joins the snapshot only because the same `/permission` switch changes it and leaving its system section would retain the cache defect.
+
+**Keep Candidate A after the cache-safe move.** Rejected by the neutral real-provider task: the model returned a pure text refusal and made no tool call despite the existing bash attempt guidance. Candidate B states no escalation mechanics itself; it tells only families whose tools actually advertise a wider retry not to infer impossibility from the standing label, then delegates denial and escalation behavior back to those tool owners.
 
 **Keep sandbox mode absent because a standing mode label once caused preemptive refusal.** Rejected because a fresh Web request otherwise exposes mutation tools while withholding their standing policy, producing false capability claims before the first operation. The earlier live measurement remains a required counter-test: five of twelve turns ended without a tool call under `Bash commands run under the "read-only" file sandbox.` The committed tool-owned attempt guidance postdates that measurement, so the replacement is selected through a new positive-control experiment under the current tool contract rather than assuming the old and current conditions match.
 
-**A separate model-context package.** Rejected because Cordis services can observe current runtime contributions directly, while approval and plan policy sections already live with their owners. A new package would add a shallow composition seam and documentation/gate surface for one internal adapter.
+**A separate model-context package.** Rejected because Cordis services can observe current runtime contributions directly and the existing assembly service can order them. A new package would add a shallow composition seam and documentation/gate surface around the same request boundary.
 
 **Enumerate writable temporary roots.** Rejected because the backend is selected later at `confine()`: bwrap, Landlock, Seatbelt, and the in-process filesystem fence do not grant one common temporary-path set. Host-specific paths in a standing request would be both unstable and overclaimed.
 
 ## Consequences
 
-A model can answer what registered file operations the standing mode governs before probing a tool, and the next request after `/permission` reflects the committed mode. This adds a small dynamic system section and intentionally invalidates the request prefix when policy or enforcing-family composition changes; unchanged state remains cache-stable. The statement is guidance, not an enforcement guard: runtime safety still comes from the registered filesystem, one-shot bash, and terminal backends consuming the same resolved policy.
+A model can answer what registered file operations the standing mode governs before probing a tool, and the next request after `/permission` reflects the committed mode. The stable system prompt no longer changes for sandbox or approval state; a changed full context snapshot is append-only after retained history, and unchanged state adds no message. Older snapshots remain in history but are explicitly superseded by the latest full snapshot. The statement is guidance, not an enforcement guard: runtime safety still comes from the registered filesystem, one-shot bash, and terminal backends consuming the same resolved policy.
 
-Focused tests pin all modes, family combinations, contribution disposal, canonical roots, switch timing, and byte stability across different `TMPDIR` values. Keyless assembled snapshots pin the request header through real Loader compositions, including all three families. Keyless replay owns the neutral denial-to-escalation trajectory; it is a structural regression proof, not wording-selection evidence.
+Focused tests pin all modes, family combinations, contribution disposal, canonical roots, switch timing, context ordering, clearing, stable request headers, and byte stability across different `TMPDIR` values. Keyless assembled snapshots pin the durable context message through real Loader compositions, including all three families. Keyless replay owns the neutral denial-to-escalation trajectory; it is a structural regression proof, not wording-selection evidence.

.agents/notes/implemented/feature/2026-07-30-current-sandbox-policy-context.zh.md

@@ -10,15 +10,19 @@ Status: implemented
 
 ## 决策
 
-`dsh-sandbox-policy` 负责解析模式与工作区根目录，并注册一个 `sandbox:policy` 系统提示词段落。每次 agent（智能体）请求都通过 `ctx.sandboxPolicy.resolve({ session })` 直接解析当前会话；不存在拒绝历史扫描、差量叙述器或内存中的「上次告知」状态。
+`dsh-sandbox-policy` 负责解析模式与工作区根目录，并注册一项缓存安全的 `sandbox:policy` 上下文贡献。每次 agent（智能体）请求都通过 `ctx.sandboxPolicy.resolve({ session })` 直接解析当前会话；不存在拒绝历史扫描或进程本地的「上次告知」状态。
 
-强制执行后端会向策略归属方注册可独立释放的 `filesystem`、`bash` 或 `terminal` 家族贡献。该段落只按规范顺序列出已注册家族，没有家族时为空。这是当前需求，而不是未来扩展：已交付的 headless inheritance 组合将沙箱化文件系统工具与不受围栏约束的一次性 bash 结合，而 persistent-tools 组合则包含沙箱化文件系统工具与终端命令，却没有沙箱化的一次性 bash 执行器。笼统声明在这两种组合中都会失实。
+强制执行后端会向策略归属方注册可独立释放的 `filesystem`、`bash` 或 `terminal` 家族贡献。该贡献只按规范顺序列出已注册家族，没有家族时为空。这是当前需求，而不是未来扩展：已交付的 headless inheritance 组合将沙箱化文件系统工具与不受围栏约束的一次性 bash 结合，而 persistent-tools 组合则包含沙箱化文件系统工具与终端命令，却没有沙箱化的一次性 bash 执行器。笼统声明在这两种组合中都会失实。
 
-该段落只说明每个已注册家族的所有强制执行方言所共有的事实。`read-only` 表明这些操作无法修改文件。`workspace-write` 用非排他措辞说明规范化的会话工作区，并概述某些平台临时区域可能也可写，而不逐一列举。`danger-full-access` 表明 DSH 文件沙箱不会限制这些操作。后端选择的临时路径、`/dev/null`、runner 就绪状态和其他策略领域都不会出现，因为 `resolve()` 无法在请求组装时确定它们。
+该贡献只说明每个已注册家族的所有强制执行方言所共有的事实。`read-only` 表明这些操作在常驻模式下无法修改文件。只有真实 schema 与执行路径提供经批准的更宽松模式重试时，工具归属方才会另行注册对应家族；只读文本只对既受强制执行又可升权的家族交集添加反预防性拒绝引导。`workspace-write` 用非排他措辞说明规范化的会话工作区，并概述某些平台临时区域可能也可写，而不逐一列举。`danger-full-access` 表明 DSH 文件沙箱不会限制这些操作。后端选择的临时路径、`/dev/null`、runner 就绪状态和其他策略领域都不会出现，因为 `resolve()` 无法在请求组装时确定它们。
 
-提供方在正常请求组装期间运行：此时 `/permission` 切换已经提交既有 `sandbox/mode` 事件，`request/header` 尚未记录。因此，渲染后的系统文本就是模型所见确切事实的持久化重建结果。会话状态不变时，重复组装会产生完全相同的字节；恢复与回放会折叠同一条持久模式事件和不可变的 `SessionHeader.cwd`，无需追赶状态。
+现有 `dsh-system-prompt` 组装在稳定系统段与工具 schema 之外，还包含有序的动态上下文。组装一个步骤后，agent loop（智能体循环）会将所有活动上下文渲染成一份带显式取代声明的完整快照。仅当不存在保留快照、字节发生变化、压缩（compaction）移除了保留消息，或最后一项贡献消失而需要一份清除快照时，它才会追加一条带来源的 `user/message`。快照追加在现有历史之后、`step/start` 之前，因此策略变化时仍会保留此前的系统与对话缓存前缀。会话事件本身可以重建确切的模型输入；只有策略上下文变化时，`request/header` 仍逐字节相同。
 
-归属范围保持收敛。批准策略仍由独立的 `approval:policy` 段落负责，计划模式仍由 `plan:policy` 负责，工具插件也继续负责各自的 schema，以及尝试、拒绝与升级引导。提示词负责说明常驻策略；文件系统、一次性 bash 与终端后端仍是强制执行边界。
+归属范围保持收敛。批准策略会将其完整的当前 `ask` 或 `never` 事实贡献给同一份完整快照；只迁移沙箱无法保留缓存，因为 `/permission` 会同时改变两方。计划模式仍由 `plan:policy` 负责，工具插件也继续负责各自的 schema，以及尝试、拒绝与升级引导。上下文负责说明常驻策略；文件系统、一次性 bash 与终端后端仍是强制执行边界。
+
+缓存决策依据当前源码，而不只依靠类比。Codex 将权限建模为 developer 角色的 `WorldState` 段并保存其指纹；只有状态变化或保留的历史丢失该片段时才发出它，同时记录快照转换。Hermes 在会话期间保持系统提示词不变，并明确将不断变化的 skill（技能）、模型与语音通知前置到下一条用户消息，以免提示词缓存失效。Pi 没有可比的内置沙箱状态，Claude Code 当前的原生实现也无法公开检视；不过，Anthropic 的公开缓存指南仍将不断变化的逐请求上下文放在稳定缓存前缀之后。
+
+先前接入真实提供方的 Web fixture（测试前置数据）量化了系统段版本的缺陷。首次 `danger-full-access` 和 `workspace-write` 请求分别只有 256 个缓存读取 token，而未缓存输入 token 为 14,691 和 14,782 个。相同策略下的后续步骤报告约 14.7k–15.5k 个缓存读取 token。只移动沙箱语句无法修复这些未命中，因为同一次 preset 切换还会改写批准策略系统段。
 
 ## 措辞证据
 
@@ -26,24 +30,32 @@ Status: implemented
 
 因此，阳性对照未通过预先登记的灵敏度门槛。Candidate A 与 B 均未运行，本实验不选择也不验证当前措辞。它只说明先前十二次中五次的结果无法在本任务与当前工具引导下复现；在声明模型行为率之前，需要更强的阳性对照或不同的任务分布。下述确定性测试只证明请求构造与回放真实一致。
 
+随后，缓存安全交付重做针对中性 Web 任务 `Create the relative path policy-neutral.txt ...` 提供了一次独立的非统计验收对比；它不取代预先登记的十二会话实验。Candidate A 的绝对化只读声明导致模型以纯文本拒绝，工具调用为零。Candidate B 只针对受强制执行、且其工具公开升权能力的家族增加一句按组合条件化的文案。随后一次全新的真实提供方运行先发出普通 `write`，观察到只读拒绝，再在同一轮次用 `sandbox_permissions: "workspace-write"` 重试同一操作，获得批准、读回文件并核验确切内容。它没有进行推测性升权。在权限切换和四个变更步骤中，每个请求的缓存读取为 14,848–15,872 个 token，未缓存输入为 59–251 个 token，直接证明了稳定前缀的收益。
+
 ## 曾考虑的替代方案
 
 **仅叙述模式变更。** 不予采用，因为这会让新会话不了解策略，并把首次被拒绝的操作变成策略发现机制。如果可以直接渲染当前状态，也就无需额外定义基线。
 
 **扫描拒绝历史或记住上次叙述的模式。** 不予采用，因为拒绝事件描述的是尝试过的操作，而不是权威的当前状态；进程本地的簿记也无法跨恢复保留。归属方可以在每次请求时直接折叠持久策略。
 
-**通用运行时事实注册表。** 不予采用，因为现有系统提示词注册表已经会在请求时使用当前 agent 评估归属方提供的段落。单一策略归属方不存在需要另一个包或注册表来承载的跨领域不变式。
+**把当前策略放入动态系统段。** 不予采用，因为真实提供方证据显示，首次权限切换后缓存读取降至 256 个 token，而约 14.7k 个输入 token 未命中缓存。DeepSeek 匹配完整前缀；改变第一条 wire 消息会阻止复用更长的系统与历史前缀。
 
-**在该段落中重复工具 schema，或批准与计划引导。** 不予采用，因为这些接口已有各自归属方和独立生命周期。重复内容会造成相互矛盾的请求前缀，并扩大缓存失效范围。
+**由每个策略归属方独立调用 `agent.inject()`。** 不予采用，因为同级监听器的顺序会决定模型所见顺序，分开的消息可能暴露不匹配的中间快照，并且每个归属方都需要各自扫描压缩后的保留状态。现有组装归属方可以对贡献排序，并具体化一份原子化的完整快照。
+
+**通用运行时事实包。** 不予采用，因为现有系统提示词组装已经拥有段、schema、变量、作用域和权威的逐步骤 waterfall（瀑布式事件）。为该归属方增加有序上下文，无需新增包或第二个注册表服务。
+
+**在上下文中重复工具 schema 或计划引导。** 不予采用，因为这些接口已有各自归属方和独立生命周期。批准的当前状态加入快照，仅仅是因为同一个 `/permission` 切换会改变它，而把它留在系统段会保留缓存缺陷。
+
+**缓存安全迁移后仍保留 Candidate A。** 不予采用，因为中性的真实提供方任务中，尽管已有 bash 尝试引导，模型仍以纯文本拒绝，且没有调用工具。Candidate B 本身不说明任何升权机制；它只针对那些工具确实公开更宽松模式重试的家族，说明不能从常驻标签推断操作不可能完成，然后把拒绝与升权行为交还给这些工具归属方。
 
 **继续省略沙箱模式，因为常驻模式标签曾引发预防性拒绝。** 不予采用，因为新的 Web 请求否则会暴露变更工具，却隐去这些工具的常驻策略，导致模型在首次操作前错误声称自身能力。先前的线上测量仍是必须执行的反证测试：使用 `Bash commands run under the "read-only" file sandbox.` 时，十二个轮次中有五个没有调用工具。已提交的工具归属方尝试引导晚于该测量，因此应通过当前工具契约下的新阳性对照实验选择替代文案，而不能假设旧条件与当前条件相同。
 
-**独立的模型上下文包。** 不予采用，因为 Cordis 服务可以直接观察当前运行时贡献，而批准与计划策略段落也已经与各自归属方放在一起。新包会为了一个内部适配器引入浅层组合 seam 和额外的文档／门禁表面。
+**独立的模型上下文包。** 不予采用，因为 Cordis 服务可以直接观察当前运行时贡献，现有组装服务也可以对它们排序。新包只会围绕同一个请求边界引入浅层组合 seam 和额外的文档／门禁表面。
 
 **枚举可写临时根目录。** 不予采用，因为后端要到稍后的 `confine()` 才会选定：bwrap、Landlock、Seatbelt 和进程内文件系统围栏并不授予一套共同的临时路径。常驻请求中的主机特定路径既不稳定，也会作出过度承诺。
 
 ## 后果
 
-模型可以在试探工具前回答常驻模式管辖哪些已注册文件操作，且 `/permission` 后的下一个请求会反映已提交的模式。这会增加一个小型动态系统段落，并在策略或强制执行家族组合变化时有意使请求前缀缓存失效；状态不变时仍保持缓存稳定。该声明是引导，而不是强制执行护栏：运行时安全仍来自已注册的文件系统、一次性 bash 与终端后端消费同一项解析完成的策略。
+模型可以在试探工具前回答常驻模式管辖哪些已注册文件操作，且 `/permission` 后的下一个请求会反映已提交的模式。稳定的系统提示词不再随沙箱或批准状态变化；变化后的完整上下文快照会在保留的历史之后仅追加，状态不变时不增加消息。较旧的快照仍保留在历史中，但最新的完整快照会明确取代它们。该声明是引导，而不是强制执行护栏：运行时安全仍来自已注册的文件系统、一次性 bash 与终端后端消费同一项解析完成的策略。
 
-聚焦测试固定了所有模式、家族组合、贡献释放、规范化根目录、切换时机，以及不同 `TMPDIR` 值下的字节稳定性。无密钥的组装快照通过真实 Loader 组合固定请求 header，包括全部三个家族。无密钥回放负责固定中性的拒绝到升级轨迹；它是结构回归证明，而不是措辞选型证据。
+聚焦测试固定了所有模式、家族组合、贡献释放、规范化根目录、切换时机、上下文顺序、清除、稳定的请求 header，以及不同 `TMPDIR` 值下的字节稳定性。无密钥的组装快照通过真实 Loader 组合固定持久上下文消息，包括全部三个家族。无密钥回放负责固定中性的拒绝到升级轨迹；它是结构回归证明，而不是措辞选型证据。