@dsh-cove/dsh-web-all 0.0.0-stage → 0.4.4

package/presets/code-vibe/skills/dsh-prose-standard/references/examples.md

@@ -0,0 +1,169 @@
+# Distilled prose examples
+
+Use these examples to identify the governing principle, not as text templates. “Balanced” preserves every load-bearing proposition with the least explanation needed at that location.
+
+## Preserve every factual clause
+
+**Original:** “The coordinator carefully serializes writes per session, flushes buffered events before disposal resolves, and reports backend failures to the caller.”
+
+**Over-trimmed:** “The coordinator serializes persistence.”
+
+**Balanced:** “The coordinator serializes writes per session, flushes buffered events before disposal resolves, and reports backend failures to the caller.”
+
+Remove decoration and repetition, not propositions. Actor, per-session scope, disposal ordering, and failure visibility are separate facts.
+
+## Explicit skill scope is functional
+
+**Over-trimmed:** “Read the sources and use judgment.”
+
+**Balanced:** “This skill is guidance, not a complete checklist. Use judgment beyond the named checks; documented requirements still apply.”
+
+**Over-detailed:** Several paragraphs defending why lists cannot replace independent reasoning.
+
+Keep the explicit limitation because it changes how an agent applies the workflow. Trim repeated persuasion, not the guardrail.
+
+## A cookbook keeps action and verification
+
+**Over-trimmed:** “Add tests for the tool.”
+
+**Balanced:** “Test registration and disposal at unit level, exercise the tool through the real loader path, and add a snapshot when its rendered output changes. Verify the assertion observes the external result rather than the model's report.”
+
+**Over-detailed:** A walkthrough of every fixture file and assertion already visible in the example code.
+
+Keep the test tiers, required action, real entry path, and observable verification. Remove fixture narration.
+
+## Preserve ownership and timing
+
+**Over-trimmed:** “Provider work is cancelled during teardown.”
+
+**Balanced:** “The runtime requests provider cancellation before releasing the child scope; the provider remains responsible for joining its workers before disposal resolves.”
+
+**Over-detailed:** A chronological account of every promise and callback used to implement teardown.
+
+The actor, ordering, point where ownership changes, and completion guarantee are separate factual clauses.
+
+## Event JSDoc preserves boundary timing
+
+**Over-trimmed:** “Composes and caches the session prefix.”
+
+**Balanced:** “Composes the session prefix once before the first pre-step and model request. Listener appends join the current request, and pre-step pressure accounting receives the composed prefix.”
+
+**Over-detailed:** A walkthrough of the loop helpers, cache fields, and promise callbacks that implement the ordering.
+
+Event order and its current-request consequence are caller-visible behavior, not implementation narration.
+
+## Orient complicated code without narrating it
+
+**Over-trimmed:** “Worker realm support.”
+
+**Balanced:** “Owns the worker realm and its host bridge. Realm initialization is single-shot; disposal terminates the worker and rejects later calls. See the worker-isolation issue or pull request for the protocol rationale.”
+
+**Over-detailed:** A paragraph-by-paragraph preview of the classes and helper functions below.
+
+Keep the module's role, dependencies, responsibilities, and non-obvious lifecycle behavior. Link architecture rationale and let the code show local control flow.
+
+## Public JSDoc includes failures
+
+**Over-trimmed:** “Returns the realm global.”
+
+**Balanced:** “Returns the initialized realm global. Throws if initialization has not completed or the realm has already been disposed.”
+
+**Over-detailed:** The internal state-machine branches and exact helper calls that lead to each throw.
+
+Throws and state preconditions are caller-visible contract facts.
+
+## Keep a concise implementation mapping
+
+**Over-trimmed:** “Search provider backed by an external API.”
+
+**Balanced:** “Maps each provider result to the shared search-result fields, preserving the title, URL, and text while omitting provider-only ranking metadata.”
+
+**Over-detailed:** A field-by-field restatement of the mapping code, including fields with identical names and obvious assignments.
+
+Keep mapping details that explain where an adapter drops or changes information.
+
+## Link rationale while keeping the local contract
+
+**Over-trimmed:** “Disposal is documented in the lifecycle issue or pull request.”
+
+**Balanced:** “Disposal aborts the run and waits for provider quiescence. See the lifecycle issue or pull request for ownership and race handling.”
+
+**Over-detailed:** Repeating the issue or pull request's promise choreography and rejected ownership models beside every disposer.
+
+Keep the behavior and completion guarantee where callers need them. Link aggressively for the algorithm and rationale; a link cannot replace the local contract.
+
+## Implemented issues and pull requests retain verification contracts
+
+**Over-trimmed:** Deleting the entire Testing section because the issue or pull request has already shipped.
+
+**Balanced:** “Unit tests cover cancellation before and after publication, disposal quiescence, and provider reload. A built-entry smoke covers the real loader path; snapshot coverage is deferred because the transport is process-specific.”
+
+**Over-detailed:** A file-by-file walkthrough of fixtures and assertions with no additional behavioral distinction.
+
+Remove migration tasks and test narration. Keep the tiers, behaviors they pin, real entry path, and named coverage gaps.
+
+## A security boundary may need one concrete example
+
+**Over-trimmed:** “Mounted plugins share the host's authority.”
+
+**Balanced:** “Mounted plugins share the host's authority; for example, access to `ctx.shell` permits commands with the host executor's privileges.”
+
+**Over-detailed:** A list of every service a plugin could misuse and every hypothetical exploit.
+
+Keep one example when it makes an otherwise abstract security limit operationally clear.
+
+## Delete reasoning transcripts entirely
+
+**Over-detailed:** “First the loop checks whether the value is absent. If it is absent, the next branch returns early. Otherwise it continues, which is why the final assertion is safe.”
+
+**Balanced:** No comment when the code already expresses those branches. If the early return protects a non-obvious invariant, state only that invariant.
+
+Do not compress a reasoning transcript into shorter narration; remove it.
+
+## Configuration comments explain what the tree cannot
+
+**Over-detailed:** “This entry loads the local filesystem provider, followed by the policy plugin, followed by the read, write, and edit tools,” when the adjacent entries already show that order.
+
+**Balanced:** “Load policy before the model-facing tools so their write and edit calls pass through the read-before-mutation gate.”
+
+Keep the consequence of order, a surprising scope rule, or a security boundary. Let the configuration show its own inventory.
+
+## Do not trim for word count alone
+
+**Current:** “The adapter converts provider errors into the shared error type so callers can handle authentication, rate-limit, and transient failures uniformly.”
+
+**Shorter but worse:** “The adapter normalizes provider errors.”
+
+**Balanced decision:** Keep the current sentence unless a link or surrounding contract already lists the failure categories. The shorter version loses the consequence and distinctions without improving structure.
+
+## Model-visible text follows ownership
+
+**Over-trimmed:** “The tool returns errors when a call fails.”
+
+**Over-detailed:** Copying another package's schema and renderer strings into this backend's README.
+
+**Balanced:** Quote stable prompt, result, and error text owned by this package. Link the generated tool catalog for schemas and the consumer README for text another package owns; state only this package's conditions or deltas locally.
+
+Wording that reaches a model is behavior, but duplication still drifts. Exactness belongs at the owner.
+
+A prose-only audit may identify suspect wording but must not silently change it when no owning runnable snapshot exists. Leave it unchanged and report the deferral, or expand the authorized change to include the owner scenario and required GUI evidence.
+
+## Generated summaries must stand alone
+
+**Over-trimmed:** “Approval request and policy service.” The owner explains policy order and audit logging later, but the catalog exports only its first sentence.
+
+**Over-detailed:** Moving the service's full lifecycle and prompt-notice behavior into the extracted sentence.
+
+**Balanced:** “Approval service that applies session policy before answerers and logs every ask/outcome pair to the requesting session.” Keep non-catalog detail in later sentences.
+
+Know what the generator extracts. That fragment must preserve the contract needed on its generated output.
+
+## Limitations are contracts, not debt inventories
+
+**Over-trimmed:** Omitting a process-lifetime cache that makes configuration changes require plugin reload.
+
+**Over-detailed:** Listing private helper cleanup and unused test-only accessors with no caller or maintainer consequence.
+
+**Balanced:** “Provider selection is cached for the plugin lifetime; installing or repairing a provider requires reload.” Keep ordinary cleanup in its TODO or issue or pull request.
+
+Retain gaps and non-obvious constraints that affect use or safe maintenance. A package README is not a backlog dump.

package/presets/code-vibe/skills/dsh-speed-up-perf/SKILL.md

@@ -0,0 +1,94 @@
+---
+name: dsh-speed-up-perf
+description: 性能测量/基准纪律； 'Use when investigating or optimizing DeepSeek Harness performance, designing realistic synthetic benchmarks or CI performance gates, profiling long Sessions or Web responsiveness, or turning performance PR evidence into measured behavior-preserving fixes.'
+
+<!-- provenance: ported from deepseek-ai/deepseek-harness/.agents/skills/dsh-speed-up-perf @ 00102833dfaee1da9f48a3a8eae9d34005a75218 (MIT License, Copyright (c) 2026 DeepSeek). Content adapted for self-contained operation outside the upstream repo. -->
+---
+
+# Speed Up DeepSeek Harness
+
+Turn a broad “make it faster” request into reproducible user-path measurements and small, evidence-backed fixes. This is guidance, not a quota or a script: survey broadly, follow measured cost, and reject attractive changes that do not improve the workload users actually run.
+
+## Establish scope and current authority
+
+Read the repository constitution（AGENTS.md/architecture/testing policy/defensive patterns——源自 DeepSeek Harness）and the affected packages’ instructions and decision records. Use [CI test reliability](../dsh-ci-test-reliability/SKILL.md) for processes, clocks, browser tests, and asynchronous cleanup.
+
+Agree on the user-visible endpoint, workload range, resource constraints, acceptable minor behavior differences, and stopping rule. Keep backend and browser end-to-end measurements separate: a fast history iterator or Client fold does not prove fast transport, paint, scrolling, or input response. Exclude model/network latency when measuring local overhead, and state that exclusion rather than calling the result complete product latency.
+
+Inspect the exact current base, not just the running checkout. Study final merged diffs, owning source, tests, and resolved review threads; a PR body can describe an abandoned implementation. Separate merged, closed-unmerged, superseded, estimated, and newly measured evidence. The performance workflow decision（上游决策档案）supply historical leads, not authority to reintroduce their implementations.
+
+## Survey user paths, then rank candidates
+
+Delegate independent domains when breadth helps; require measurements and production call sites, not guesses. Useful domains include:
+
+- Cold profile startup, first historical read, current-generation reopen, and writable resume.
+- Many-turn and tool-heavy history, large individual messages/results, child Session listing, and repeated navigation among Sessions.
+- Initial history transport and fold, first usable browser paint, older-page loading, scrolling, tool expansion, and inactive-view activation.
+- Live streaming and reconnect, including a long active attempt, interleaved tool work, settlement, cancellation, and teardown.
+
+Vary independent cost drivers: bytes, durable events, compact records, raw deltas, turns, tools, children, and visible DOM nodes are different quantities. Do not call a large count of tiny identical messages “realistic” without checking which user operation it stresses. Include typical and tail workloads, but avoid a combinatorial matrix with no decision value.
+
+Rank candidates by observed user latency, CPU/allocations, retained memory, occurrence, and confidence. For each, name the production consumer, the repeated work, the expected complexity, the smallest falsifiable intervention, and the behavior that must remain stable. A suspicious loop, unused cache, or large file alone is not evidence of a bottleneck.
+
+## Build realistic synthetic benchmarks first
+
+Follow benchmark discipline（基准纪律：扩展既有必需通道而非另建校准/报告设施——源自 DeepSeek Harness 的 benchmarks/AGENTS.md 与性能门禁决策档案）。 Extend the existing required lane rather than creating competing calibration or reporting infrastructure. Package-local diagnostics remain beside their owner; cross-package required cases live under the measured user path in `benchmarks/`.
+
+If the user authorizes local corpus inspection, extract only aggregate workload characteristics. Never copy prompts, outputs, paths, identities, IDs, credentials, recordings, or recognizable snippets into fixtures, logs, screenshots, PRs, or artifacts. Generate fixed inputs from reviewed constants; no benchmark depends on the user’s home, ambient repository, network service, or private data.
+
+Before implementation, record a measurement card:
+
+| Field | Required decision |
+|---|---|
+| User operation | Exact action and externally observable completion condition |
+| Workload | Fixed dimensions, distributions, construction seed/constants, and why they exercise ordinary and tail use |
+| Entry path | Production calls/composition and built artifacts; mocked external boundaries |
+| Clock | Included setup, cold/warm state, timing start/end, and excluded costs |
+| Memory | Reachable endpoint objects, baseline, GC policy, retained versus transient limits |
+| Verdict | Raw samples, chosen aggregate, calibrated absolute/ratio/memory limits, and negative control |
+| Behavior | Owning functional tests/snapshots and permitted minor differences |
+
+Measure built JavaScript under plain Node for CPU workers; source-loader overhead and module resolution are not the shipped path. Browser cases use built product assets and the supported `dsh` profile through the existing test harness. Do not add a production export solely for measurement or copy the algorithm into a “benchmark implementation.”
+
+Use fresh children and private temporary roots for cold/process-memory samples. Warm samples explicitly retain the intended cache; never let fixture setup secretly warm a cold scenario. Keep the same input, validations, completion condition, and reachable output on both sides. A parse-and-discard baseline is not comparable with validated retained history.
+
+Report all samples and the aggregate that decides the result. For the Node lane, use the existing shared time calibration and reviewed variance headroom; do not scale bytes, counts, or dimensionless ratios by CPU speed. Keep manual browser diagnostics threshold-free. A required browser performance case needs an explicit lane decision and repeated measurements on its actual CI browser/runner before adopting timing budgets; the Node machine multiplier alone is not browser calibration. Budgets are source constants, not environment overrides. Serialize measured work against other owned CPU-heavy jobs; measure reference and candidate under comparable conditions. Do not widen a budget or select a lucky run to hide a regression.
+
+Measure end-to-end latency independently from component phases. Track retained memory with intended objects still reachable, and transient pressure separately through constrained-heap completion or an appropriate peak measurement. Faster execution with unbounded retention is not an automatic win.
+
+For browser responsiveness, use real browser input and observe the resulting UI update. Include the final stall in frame/input measurements, distinguish scheduled timers from actual input, and bound synthetic producers so catch-up bursts do not invent a different workload. State whether first paint, scrolling, paging, live updates, and activated-but-hidden views are covered. Node folds, fake DOMs, and custom heartbeat events alone cannot establish browser responsiveness.
+
+## Prove the regression, then remove work
+
+Run the unoptimized workload before changing production code. Save the command, revision, runtime/platform, fixture dimensions, raw measurements, and verdict. Reduce a failing scenario until it still exercises the real bottleneck, then rank falsifiable hypotheses before patching. Use profiles, allocation samples, work counts, or phase timings to distinguish them.
+
+Common patterns worth testing, not automatic prescriptions:
+
+- Keep compact representations compact through downstream readers; avoid per-delta objects when the consumer needs settled content or one aggregate.
+- Remove duplicate parsing, copying, freezing, and validation only after identifying the actual ownership and trust transition. Typed same-process borrowing is not permission to weaken durable or wire parsing.
+- Stream artifact transformations and bound intermediate state rather than retaining every generation. Include publication, verification, and writable-readiness obligations where the user operation requires them.
+- Separate read-only preparation from write/publication work without moving awaited work past a correctness-required endpoint.
+- Defer inactive-view and collapsed-detail work; measure first activation and retained state too. Deferral is not deletion, and viewport highlighting is not full virtualization.
+- Stabilize identities and narrow subscriptions so one changed node does not invalidate an entire history; preserve update ordering and immediate-event behavior.
+- Prefer a suitable data structure to repeated shifting, scanning, or rebuilding. Measure the whole consumer path, not just the isolated container operation.
+- Use revision-keyed reuse or singleflight only with explicit invalidation, bounded retention, independent waiter cancellation, and disposal ownership. Avoid caching expanded representations merely to make repeated benchmarks look fast.
+
+Change one causal factor at a time. Re-run both the focused scenario and its end-to-end parent. Require a negative control: the tightened assertion fails on the original implementation or a controlled reintroduction of the targeted cost. A threshold so generous that the regression passes is not protection; a budget below a verified noise floor is not reliable either.
+
+## Preserve behavior and resource ownership
+
+Performance measurements complement functional evidence; they do not replace it. Run or add the narrow owning tests for output, ordering, paging, stream indexes, errors, cancellation, concurrency, and disposal as applicable. Preserve model-visible/logged equivalence, released-generation immutability, atomic publication, required validation, and writable readiness. Do not silently truncate history, skip tool results, disable invariants, or change lifecycle semantics to reach a number.
+
+State any deliberate minor visible difference and verify it through the owning keyless snapshot. For a product-visible GUI change, include the required browser evidence/GIF. Keep functional expectations independent of benchmark internals; benchmark assertions need enough evidence to reach the real endpoint, not a second semantic test suite.
+
+Reject an optimization when gains disappear end-to-end, a typical workload regresses materially, complexity outweighs a small gain, or cancellation/retention/durability cannot be explained and tested. Record the rejected hypothesis briefly instead of expanding scope to justify it.
+
+## Deliver a bounded, reviewable result
+
+Use decision-record rules（耐久理由/备选/校准/排除/剩余风险——源自 DeepSeek Harness note 体系）for durable rationale, alternatives, calibration, exclusions, and remaining risks. Check relevant notes for supersession without turning performance work into a corpus-wide prose cleanup. Keep the reusable procedure here and scenario-specific truth with its benchmark or package owner.
+
+When the task requests stacked PRs, choose layers before editing and use official GitHub stacks and separate worktrees. Keep each layer mergeable: benchmark infrastructure can protect the measured baseline; the optimization layer carries its fix, functional coverage, and tighter budget. Independent bottlenecks may use separate stacks. Fix a finding in its owning layer before propagating upward.
+
+Apply pre-push checks（最小证据集纪律）, report only executed evidence, and inspect CI rather than assuming local timing proves runner stability. After marking ready, evaluate review findings against code and executable evidence; reply with the reason or fix and resolve addressed threads. Do not dismiss a report merely because it came from a bot.
+
+Summarize each result as: workload → before/after absolute values and ratio → endpoint and memory semantics → behavior evidence → negative control → exact checks → exclusions. Separate author-reported historical numbers, fresh local measurements, and CI evidence. Stop at the agreed scenario/fix scope; retain a short ranked follow-up list instead of chasing unrelated opportunities.

package/presets/code-vibe/skills/dsh-trim-cot-leakage/SKILL.md

@@ -0,0 +1,47 @@
+---
+name: dsh-trim-cot-leakage
+description: 思维链泄漏清理/CoT 残留； Use when auditing or fixing prose that reads like a leaked reasoning transcript — dead design-session citations such as (decision N), audit item codes, or §N of uncommitted drafts; change narration such as "used to", "no longer", "this cut"; stack or review vantage ("a later PR in this stack", "rejected in review"); reviewer-addressed justifications; control-flow narration; or hedged planning residue in comments, JSDoc, docs, or issues and pull requests.
+
+<!-- provenance: ported from deepseek-ai/deepseek-harness/.agents/skills/dsh-trim-cot-leakage @ 00102833dfaee1da9f48a3a8eae9d34005a75218 (MIT License, Copyright (c) 2026 DeepSeek). Content adapted for self-contained operation outside the upstream repo. -->
+---
+
+# Trimming Chain-of-Thought Leakage
+
+Chain-of-thought leakage is prose whose vantage is the authoring session rather than the repository: it cites artifacts only that session could see, narrates the change instead of the state, or argues with a reviewer who has left. The fix is never deletion alone when a passage carries factual clauses — restate each so it stands at HEAD, then delete the transcript around it; a passage carrying none (an audit code, control-flow narration) is deleted outright. **REQUIRED BACKGROUND:** [dsh-prose-standard](../dsh-prose-standard/SKILL.md) owns the complete-proposition rule this skill applies; the committed-artifact-citations rule（引用必须指向已提交工件）owns the citation rule's rationale. It is guidance, not a script.
+
+## The one test
+
+For every suspect passage ask: **could a reader at HEAD, with no access to any session transcript, PR thread, or uncommitted draft, resolve every reference and verify every claim?** If no, restate the surviving facts from the repository's vantage and delete the rest. If yes, it is not leakage, however historical it sounds — but resolvability only clears this skill's bar: on current-state surfaces (READMEs, docs, JSDoc) a resolvable change story is still change narration, and class 3 routes it to its sanctioned home.
+
+## Taxonomy
+
+1. **Dead design-session citations** — `(decision 7)`, `(audit C2)`, `design §4.7`, `plan §1.4`, phase labels (`T4`, `W3`, `P-I`), "the design ledger", "(B ruling)". If the decision has a committed owner, cite it by name and path; otherwise delete the citation and restate its factual clause to stand alone.
+2. **Stack and PR vantage** — "a later PR in this stack", "this PR adds", "the previous commit". State the shipped mechanism or the extension point; deferred work moves to a `TODO` marker or an issue reference.
+3. **Change narration and version stamps** — "used to", "no longer", "the old X", and indexical stamps ("v1", "this cut", "today", "now" contrasting with a past state). State the present behavior; a fixed regression becomes a present-tense counterfactual ("without X, Y happens"), never repo history ("used to Y").
+4. **Review choreography** — "Rejected in review:", "the reviewer confirmed", draft ordinals ("v5 of this note"), round attributions. Keep the surviving decision and rationale as plain fact; delete who said it when.
+5. **Reviewer-addressed justification** — "the cast is safe — it simply…", "this is correct because…". A comment arguing its own correctness addresses a reviewer, not a maintainer. State the invariant that makes the code safe, or delete the comment if the code shows it.
+6. **Restatement and derivation transcripts** — control-flow narration ("first we X, then we Y"), test walkthroughs, proofs of obvious branches. Delete; keep only a non-obvious contract or invariant.
+7. **Hedges and planning residue** — "probably fine for now", "should be enough", deferrals with no marker. Promote to `TODO`/`FIXME` or restate as the actual bound; delete the hedge.
+8. **Authoring-language slips** — untranslated working-language fragments (端, 设计稿, `---- 私有 ----` separators) in prose whose language is otherwise English, or the reverse in a zh counterpart. Translate or delete.
+
+## What is not leakage
+
+Unaided citation passes fail in both directions by deleting durable references and keeping dead ones. Apply these keep rules as written; [examples](references/examples.md) calibrates each:
+
+- **Issue references** — `#1470`, `TODO(name):`, "issue #N owns the follow-up" resolve at HEAD; keep them on any surface, including READMEs. Do not relocate them to issues and pull requests.
+- **Merged-PR and issue citations inside issues and pull requests and postmortems** — sanctioned evidence per the documentation standard（文档分层，源自 DeepSeek Harness）'s change-story routing.
+- **Suppression justifications** — `oxlint-disable … -- reason`, coverage-ignore reasons, empty-catch explanations are required prose; fix a false reason, never delete it.
+- **Counterfactual-present regression pins** — "without X, Y happens", "a naive X would…".
+- **Measured bounds** — "(measured: 512 nests ≈ 0.15s)" calibrating a constant; the evidence word "measured" is load-bearing.
+- **Runtime old/new states** — "the old connection drains before the new one accepts" is runtime lifecycle, not change history.
+- **Historical stage names inside a note's change-story sections** — "the first cut shipped X" is current-state-safe there; indexical stamps ("this cut") stay banned everywhere.
+- **External references that resolve outside the repo by design** — standards sections (RFC 9110 §10.1.5), Figma frame names; the §-ban covers uncommitted internal drafts, not external standards or committed docs that own their §-numbering.
+- **Project voice and genre forms** — "we" as project voice; a note's Alternatives-considered section.
+
+## Workflow
+
+1. Scope and exclusions per [dsh-prose-standard](../dsh-prose-standard/SKILL.md): require an explicit scope; never touch `vendor/` or `archived third-party snapshots/`. Recorded fixtures and snapshots are derivatives, not prose targets: change the owning source or scenario and regenerate them only when an authorized behavior change requires new evidence.
+2. Audit read-only first: run the [recall batteries](references/recall-batteries.md) (with `--hidden` so `.agents/` is searched), calibrating each probe against a known positive and a near-miss negative before trusting its output, then judge every hit semantically. The batteries are probes, not the definition — each review round of the original purge found cases the batteries missed, so also read the densest prose in scope (module JSDoc, READMEs, issues and pull requests) without a pattern in hand.
+3. Fix owner-first per surface: generated catalogs → trace every consumer, fix the source JSDoc or generator template, then regenerate all derivatives; type-equivalence fences → fix the source JSDoc, then re-paste both bilingual pages（双语对须人工逐字对照一致；上游的 verify-type-equiv 门禁本环境不存在，以对照清单代替）; bilingual prose → update the counterpart minimally and re-record it through the lightweight routine（双语最小更新路径）; bilingual fences → copy the corrected verbatim block byte-for-byte into both sides per bilingual-doc rules（双语文档纪律），then re-record the pair; model- or user-visible strings → route through [dsh-prose-standard](../dsh-prose-standard/SKILL.md) and change only with owning behavior evidence, otherwise leave unchanged and report the deferral.
+4. Before deleting anything, enumerate the passage's propositions (prose-standard) and check the [overcorrection traps](references/examples.md#overcorrection-traps): trims that flip an obligation into an endorsement, promote a hypothetical to a shipped feature, delete a true fact, or drop the evidence source.
+5. Verify: re-run the batteries expecting only sanctioned keeps, this skill's own directory, and the owning note's quoted evidence; confirm every remaining citation resolves at HEAD; run the gates for touched surfaces（文档类：git diff --check + wish-programming 的 quote/模板一致性检查；双语对：人工逐字对照；本环境无 doc-sync / verify-type-equiv / verify-translation-pairing 门禁，未校验项如实报告 pending）。

package/presets/code-vibe/skills/dsh-trim-cot-leakage/references/examples.md

@@ -0,0 +1,275 @@
+# Few-shot leakage examples
+
+Distilled from the 2026-08 repo-wide purge and its review rounds. Use them to identify the governing principle, not as text templates. This file deliberately quotes leaked wording as calibration material — the [recall batteries](recall-batteries.md) exclude the skill's directory, and its wording is not a license elsewhere.
+
+## Dead citations
+
+### Decision ordinal with a committed owner
+
+**Leaked:** "Slash input resolves against the visible catalog (decision 21)."
+
+**Fixed:** "Slash input resolves against the visible catalog — the plain-text-reference decision, owned by the web input-machine note（上游决策档案）."
+
+The ordinal resolves nowhere at HEAD; the decision's name and owning note path do. Name the owning note's path at least once per file — as a link where the surface supports one — and later mentions may use the searchable name alone.
+
+### Decision ordinal without an owner
+
+**Leaked:** "The registry rejects duplicate names (decision 7: names are flat, no namespacing)."
+
+**Fixed:** "The registry rejects duplicate names; names are flat, with no namespacing."
+
+No committed artifact owns "decision 7", so the citation is deleted — but its factual clause (flat names) is restated to stand alone, not deleted with it.
+
+### Audit item codes
+
+**Leaked:** "Rendering is pure: same snapshot, same string (audit R3)."
+
+**Fixed:** "Rendering is pure: same snapshot, same string."
+
+There is no audit document in the repo; the code is pure session shorthand carrying zero propositions.
+
+### Section numbers of uncommitted drafts
+
+**Leaked:** "Layering follows the design (v2 §3.2): `src/core/` is the pure core."
+
+**Fixed:** "Layering: `src/core/` is the pure core."
+
+`§N` of a draft nobody committed is unresolvable. Contrast: "escapes per RFC 9110 §10.1.5" stays — an external standard resolves outside the repo by design, and a committed doc that owns its §-numbering may be cited by section.
+
+### Plan-phase labels
+
+**Leaked:** "`src/client/` is the shell (T4); the P-I migration owns the adapters."
+
+**Fixed:** "`src/client/` is the shell; the adapters live in `src/client/adapters/`."
+
+Phase labels index a plan that never landed. Replace the label with what the phase produced.
+
+## Stack and PR vantage
+
+### Stack position in durable prose
+
+**Leaked:** "A future remote backend implements this interface (the sandbox backend is a later PR in this stack)."
+
+**Fixed:** "A remote backend can implement this interface without changing the render layer."
+
+Durable prose cannot see the stack. Keep the extension-point contract; the pending work's home is the PR itself, a `TODO`, or an issue.
+
+### "This PR" in a README
+
+**Leaked:** "This PR adds cursor-based pagination to the session list."
+
+**Fixed:** "The session list paginates by cursor."
+
+A README outlives every PR; state the mechanism as current fact.
+
+## Change narration and version stamps
+
+### War story with a PR number
+
+**Leaked:** "Colors used to come from `--widget-*` tokens, which nothing defined, so it always rendered the fallbacks; the alias tokens fixed that (PR #88)."
+
+**Fixed:** "Colors come from the alias tokens; an undefined token renders the fallbacks."
+
+Both live facts survive — the current mechanism and the standing failure behavior — restated in the present. The bug's biography belongs to the PR and its issue or pull request.
+
+### Removal narration
+
+**Leaked:** "The `probe` field is gone with the removal cut; badges ride the generic projection pair now."
+
+**Fixed:** "Badges use the generic projection pair."
+
+Readers who never saw `probe` learn nothing from its absence. "Now" contrasting with a deleted past is a version stamp.
+
+### Fixed regression → counterfactual present
+
+**Leaked:** "This used to double-encode multibyte labels."
+
+**Fixed:** "Without the byte-length guard, multibyte labels double-encode."
+
+The regression pin survives as a present-tense counterfactual that names the guard; "used to" pins it to repo archaeology instead.
+
+### Indexical version stamps
+
+**Leaked:** "Batch rendering is synchronous this cut; the async path is roadmap work."
+
+**Fixed:** "Batch rendering is synchronous." (The deferral lives in `TODO(widget-batch):` at the call site.)
+
+"This cut" / "v1" / "today" go stale the moment they merge. A historical stage name inside an issue or pull request's change-story section ("the first cut shipped X") is current-state-safe; the indexical form never is.
+
+## Review choreography
+
+### Review verdicts as prose
+
+**Leaked:** "Rejected in review: caching the resolved spec. We keep resolution per-call."
+
+**Fixed (in an issue or pull request's Alternatives considered):** "**Caching the resolved spec.** Rejected: the spec depends on per-call cwd, so a cache keyed by request would serve stale roots."
+
+The alternatives-considered genre is the sanctioned home; the reviewer and the round are not part of the rationale.
+
+### Draft ordinals
+
+**Leaked:** "As of v5 of this note, the loader also validates manifests."
+
+**Fixed:** "The loader validates manifests."
+
+An implemented note states shipped reality; its own revision history lives in git.
+
+## Reviewer-addressed justification
+
+### Arguing a cast
+
+**Leaked:** "The cast is safe — the SDK constructed the object, it simply doesn't declare the optionals strictly enough."
+
+**Fixed:** "The SDK constructs this object with every optional populated; the declared type is looser than the runtime guarantee."
+
+State the invariant a maintainer must not break. "It simply…" is a voice answering an objection nobody at HEAD raised. If the invariant is visible in the code, delete the comment instead.
+
+### Appeal to review authority
+
+**Leaked:** "This is correct because the reviewer confirmed the wrapping order."
+
+**Fixed:** (deleted; the wrapping order is stated in the function's `@returns`.)
+
+Correctness claims cite invariants or tests, never people.
+
+## Restatement and derivation
+
+### Control-flow narration
+
+**Leaked:** "First we normalize the label, then we truncate it, then we wrap it."
+
+**Fixed:** (deleted.)
+
+The three lines below the comment say the same thing in code.
+
+### Test walkthrough
+
+**Leaked:** "This test creates a session, sends two messages, waits for the second reply, and then asserts the log has four entries."
+
+**Fixed:** "Two round-trips must produce exactly four log entries — the projection dedupes the shared prefix."
+
+Keep only the non-obvious assertion rationale; the walkthrough restates the test body.
+
+## Hedges and planning residue
+
+### Unmarked deferral
+
+**Leaked:** "Probably fine to render eagerly for now."
+
+**Fixed:** (deleted; the deferral already has its `TODO(widget-batch):` marker.)
+
+A hedge without an owner is planning residue. If no marker exists, write one (`TODO(name): coalesce per animation frame`) instead of keeping the hedge.
+
+### Vague sizing
+
+**Leaked:** "A 64 KiB buffer should be enough for most cases."
+
+**Fixed:** "64 KiB holds the largest observed frame (48 KiB) with headroom; a larger frame fails loudly in `decode`."
+
+Replace the hedge with the actual bound and the failure behavior when it is exceeded.
+
+## Authoring-language slips
+
+**Leaked:** "The renderer runs on the client 端; see the 设计稿 for spacing. ---- 私有 ----"
+
+**Fixed:** "The renderer runs on the client side; spacing follows the Figma frame `widget-badges`."
+
+Working-language fragments and session separators are transcription residue. The Figma frame name stays: an external source that resolves outside the repo by design.
+
+### Authoring-language slip inside a paired fence
+
+**Leaked in both files:** `// 更新这里 before returning` inside a verbatim code block.
+
+**Fixed in both files:** `// Update this before returning.`
+
+Correct the block once and copy that byte-exact fence into both language files. Translating the code comment differently in the Chinese counterpart breaks the pairing contract even when both comments are individually fluent.
+
+## Behavior-visible candidates
+
+**Suspect:** An exported JSDoc sentence says "available for now," and a generator copies that sentence into a model-visible catalog.
+
+**Wrong:** Rewrite only the source sentence during a prose-only purge, or hand-edit only the generated catalog.
+
+**Right:** Trace the source's generated fan-out, update the owner, regenerate every derivative, and update the owning runnable snapshot. For a GUI string, include the repository-required behavior evidence. If the authorized scope has no owning scenario, leave the wording unchanged and report the deferral.
+
+## Keeps
+
+### Issue references are durable on every surface
+
+**Keep:** "The cap applies to the complete rendered value, wrappers included (issue #1470 owns the follow-up)."
+
+An unaided pass deleted this, reasoning that issue citations belong in issues and pull requests. Wrong direction: issues resolve at HEAD from any surface, and "#N owns the follow-up" is the sanctioned home for deferred work in a README. What issues and pull requests and postmortems additionally sanction is citing *merged PRs* as evidence.
+
+### Dead name-drops are not "naming the owner"
+
+**Delete:** "Badge renderer over the widget seam (see the widget-rendering RFC)."
+
+An unaided pass kept this as "naming the owning document by topic". The test is resolvability, not form: no committed file answers to "the widget-rendering RFC", so the pointer is dead. Retarget it to the committed owner if one exists; otherwise delete it.
+
+### Suppression justifications
+
+**Keep (after fixing):** `// oxlint-disable-next-line no-non-null-assertion -- the one-element literal guarantees index 0.`
+
+The justification clause is required prose. When the stated reason is false (the original said "the loop guard above proves a frame exists" with no loop in sight), fix the reason; never delete it.
+
+### Measured bounds
+
+**Keep:** "Depth cap (measured: 512 nests ≈ 0.15s synchronous; 4096 blocks the loop)."
+
+The measurement pins the constant against uninformed retuning, and "measured" identifies data rather than a guess.
+
+### Runtime old/new is not change history
+
+**Keep:** "The old connection drains before the new one accepts."
+
+"Old" and "new" here name two live runtime objects during handover, not repository states. The change-narration ban is about repo history, not lifecycle vocabulary.
+
+### Runtime natural time is not a version stamp
+
+**Keep:** "What is today's date?"
+
+The prompt asks about the runtime clock; "today" does not contrast repository states. Because the text reaches a model, any rewrite still requires its owning behavior evidence.
+
+## Overcorrection traps
+
+Every trap below shipped in the original purge and was caught in review. Enumerate a passage's propositions before trimming it.
+
+### Flipping an obligation into an endorsement
+
+**Original:** "These direct registrations are exceptions pending migration to slots."
+
+**Overcorrected:** "These direct registrations are sanctioned exceptions."
+
+**Right:** "These direct registrations are exceptions pending migration to slots."
+
+"Pending migration" is an obligation; "sanctioned" blesses the status quo. The trim inverted the sentence's modality while shortening it.
+
+### Promoting a hypothetical to a shipped feature
+
+**Original:** "A future IPC-based shell subclasses the executor and overrides `spawn`."
+
+**Overcorrected:** "An IPC-based shell subclasses the executor and overrides `spawn`."
+
+**Right:** "A hypothetical IPC-based shell — no such shell exists — would subclass the executor and override `spawn`."
+
+Deleting the future-marker alone turns a design illustration into a claim that the class ships. Mark the hypothetical explicitly instead of just unmarking the future.
+
+### Deleting a true fact with the transcript around it
+
+**Original:** "The gate notice narrates the check order; the notice text is also what `verify-doc-typecheck` compiles against."
+
+**Overcorrected:** "…" (whole sentence deleted as narration.)
+
+**Right:** "The notice text is what `verify-doc-typecheck` compiles against."
+
+Half the sentence was narration; the other half was a load-bearing coupling. Delete clauses, not sentences, when propositions share a line.
+
+### Dropping the measurement source while keeping the number
+
+**Original:** "The 4 MiB ceiling is measured: the largest generated `py-types` module is 3.1 MiB."
+
+**Overcorrected:** "The ceiling is 4 MiB; the largest generated `py-types` module is 3.1 MiB."
+
+**Right:** keep "measured".
+
+Without "measured" the 3.1 MiB reads as a definition rather than an observation, and nobody re-measures before raising the ceiling.