fix(core): honor Shell truncation threshold setting - #9014
Conversation
github-actions
Bot
added
the
review/self-reported
E2E test reportBaselineUsing the globally installed Qwen Code 0.21.9 under the macOS seatbelt sandbox, I configured Current sourceI ran the same command through the real
The test harness executed the real child process and cleaned all temporary output afterward. Automated verification
LimitationThe repository-wide local CLI entry was not used for the after case because this checkout had unrelated stale ignored package build artifacts ( |
|
Thanks for the PR! Template looks good ✓ Problem: observed bug, not theory. The linked issue (#8922) carries a concrete reproduction — set Direction: clearly aligned — this is a documented setting that the Shell path silently ignores, so honoring it fixes a real doc-vs-behavior contradiction for users who tuned the threshold. No CHANGELOG reference upstream for this exact knob, but tool-output truncation configurability is actively maintained territory, and the area is squarely relevant. Size: core paths are touched ( Approach: the scope feels right. The interesting design point is provenance: the global default for this setting is 25,000 while Shell's own default is 30,000, so simply reading Risk: elevated — Moving on to code review. 🔍 中文说明感谢贡献！ 模板完整 ✓ 问题：已观测到的 bug，不是理论问题。关联 issue（#8922）提供了具体复现——将 方向：明确对齐——这是一个被 Shell 路径静默忽略的已文档化设置，修复它解决的是真实的"文档与行为不符"问题。上游 CHANGELOG 没有针对该配置项的直接条目，但工具输出截断的可配置性一直是活跃维护的领域。 规模：触及核心路径（ 方案：范围合理。关键设计点是配置来源（provenance）：该设置的全局默认值是 25,000，而 Shell 自身默认是 30,000，如果直接读 风险：升级—— 进入代码审查 🔍 — Qwen Code · qwen3.8-max Reviewed at |
Code reviewWorking from the stated problem, my independent proposal was exactly what this PR does: stop Shell from passing a hardcoded per-call threshold that overrides the configured one — but with one catch the PR handles correctly. The global default for What I verified statically against the code at this commit:
No correctness, security, or convention issues found. Testing evidence (this PR's own CI, read via API — no PR code executed here)The unit suite on the reviewed commit is still in flight (~30 min runtime; not polling — the finalize workflow updates the table below in place once CI settles). No red checks so far. The macOS/Windows unit jobs and the CLI integration job report "skipped" by workflow design — they only run in the merge queue (
Sandboxed verification would settle this: Real-scenario (tmux) testing: not run — this is an unattended CI pass; live behavior is exercised only by the isolated 中文说明代码审查：从问题出发，我独立给出的方案与本 PR 一致，而且它正确处理了其中的关键陷阱——该设置的全局默认值是 25,000，Shell 自身默认是 30,000，如果直接读 测试证据：来自该 PR 自身 CI 的 API 读取（此处未执行任何 PR 代码）。评审提交上的 ubuntu 单测套件仍在运行（约 30 分钟；不轮询，finalize 工作流会在 CI 结束后原地更新表格）。目前无红色检查。macOS/Windows 单测与 CLI 集成任务按工作流设计只在合并队列（merge_group）运行，PR 上显示 skipped 属预期，不是本 PR 造成的缺口。 沙箱验证建议： 真实场景（tmux）测试：未执行——这是无人值守 CI 流程；如需真实 TUI 行为验证，可由维护者触发隔离的 — Qwen Code · qwen3.8-max Reviewed at |
|
Confidence: 4/5 — clean review with zero findings; withholding the fifth point only because end-to-end proof of the behavior rests on the author's macOS run and mocked unit tests until the named This one earns its merge. The problem is real and maintainer-triaged (a documented setting silently ignored by the Shell path, P2 bug with a concrete reproduction), and the fix is the minimal correct shape — including the non-obvious part: since the setting's global default (25,000) differs from Shell's own default (30,000), honoring the value required tracking whether it was explicitly configured rather than just reading the getter. The diff does exactly that, updates both truncation boundaries through one shared helper so they can't drift, carries focused regression tests for the load-bearing case (40k output survives a 100k threshold, no marker, no spill file), and keeps every doc that described the old behavior in sync. Nothing in the diff is unrelated; if I maintain this in six months I'll thank the author, not curse them. The author's other contributions here are few and previously merged — this is merit evaluation of one PR, not volume pressure. The shell.ts path is flagged high-revert-risk in this repo's history, which is why the CI evidence is a hard precondition and the sandboxed lane is named above rather than implied. Approval deferred until CI lands green on 中文说明这个 PR 配得上合并。问题真实且已被维护者分诊（文档化的设置被 Shell 路径静默忽略，P2 bug，有具体复现）；修复是最小且正确的形态——包括不那么显眼的关键点：该设置的全局默认值（25,000）与 Shell 自身默认值（30,000）不同，因此遵循配置值需要记录"是否显式配置"，而不是直接读 getter。diff 正是这么做的：通过一个共享 helper 同时更新两个截断边界使其不会漂移，为关键场景（100k 阈值下 40k 输出完整保留、无标记、无落盘）补了聚焦的回归测试，并把所有描述旧行为的文档同步更新。diff 中没有无关改动。作者在本仓库的过往贡献数量不多且已有合并记录——这是对单个 PR 的择优评估。shell.ts 路径在本仓库历史中属高回滚风险，因此 CI 证据是批准的硬性前提，沙箱验证通道也被明确指名而非含糊带过。 批准将推迟到 CI 在该提交上全绿之后——ubuntu 单测套件仍在运行。全绿后由 finalize 工作流发布绑定该提交的批准；若出现红色检查或 head 移动，则改为标记提示。 — Qwen Code · qwen3.8-max Reviewed at |
1bbada0 to
096a3e7
Compare
|
Please do not rebase or force-push to an active PR as it invalidates existing review comments. Note for future reference, the bots always squash all changes into a single commit automatically as part of the integration. 中文请勿对活跃的 PR 执行 rebase 或 force-push，因为这会使已有的评审评论失效。另外，供日后参考：作为集成流程的一部分，机器人始终会自动将所有改动压缩（squash）为单个提交。 |
Preserve Shell's 30k default only when the global threshold is absent, and use an explicit value for both producer and scheduler truncation.
096a3e7 to
a9b2b81
Compare
qwen-code-ci-bot
left a comment
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
Reviewed. Suggestions are inline.
Not explored to full depth (tool budget reached): PR #9014 makes the Shell tool honor an explicitly configu...: none — all checks above completed; the only item left unverified is the existence of a null -stripping validation layer outside settings.ts (noted in the fin…; You are hunting for gaps the base fan-out missed in PR #9...: none — the walk is complete..
中文说明
已审查。 建议见行内评论。
未探索到全部深度（达到工具调用预算）：PR #9014 makes the Shell tool honor an explicitly configu...：none — all checks above completed; the only item left unverified is the existence of a null -stripping validation layer outside settings.ts (noted in the fin…；You are hunting for gaps the base fan-out missed in PR #9...：none — the walk is complete.。
— qwen3.8-max via Qwen Code /review (v0.21.10)
qwen-code-ci-bot
left a comment
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
Reviewed. Suggestions are inline.
Not reviewed: build-and-test — Integration Tests (CLI, No Sandbox) was skipped in CI and its suite did not run locally.
Not explored to full depth (tool budget reached): PR #9014 makes the Shell tool honor an explicitly configu...: did not verify whether the /set dialog / settings writer enforces the number type on truncateToolOutputThreshold input (concluded immaterial — the generic….
中文说明
已审查。 建议见行内评论。
未审查：build-and-test — Integration Tests (CLI, No Sandbox) was skipped in CI and its suite did not run locally。
未探索到全部深度（达到工具调用预算）：PR #9014 makes the Shell tool honor an explicitly configu...：did not verify whether the /set dialog / settings writer enforces the number type on truncateToolOutputThreshold input (concluded immaterial — the generic…。
— qwen3.8-max via Qwen Code /review (v0.21.10)
qwen-code-ci-bot
left a comment
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
Reviewed. Suggestions are inline.
Not reviewed: build-and-test — Integration Tests (CLI, No Sandbox) was skipped in CI and its suite did not run locally.
Not explored to full depth (tool budget reached): PR #9014 makes the Shell tool honor an explicitly configu...: none — all checks above completed within budget.; PR #9014 makes the Shell tool honor an explicitly configu...: full packages/core suite run (I ran only the two directly affected shell test files; config.test.ts additions were code-reviewed but not executed).; PR #9014 makes the Shell tool honor an explicitly configu...: none — all planned checks completed within budget..
中文说明
已审查。 建议见行内评论。
未审查：build-and-test — Integration Tests (CLI, No Sandbox) was skipped in CI and its suite did not run locally。
未探索到全部深度（达到工具调用预算）：PR #9014 makes the Shell tool honor an explicitly configu...：none — all checks above completed within budget.；PR #9014 makes the Shell tool honor an explicitly configu...：full packages/core suite run (I ran only the two directly affected shell test files; config.test.ts additions were code-reviewed but not executed).；PR #9014 makes the Shell tool honor an explicitly configu...：none — all planned checks completed within budget.。
— qwen3.8-max via Qwen Code /review (v0.21.10)
qwen-code-ci-bot
left a comment
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
Reviewed.
Not reviewed: build-and-test — Integration Tests (CLI, No Sandbox) was skipped in CI and its suite did not run locally.
Not explored to full depth (tool budget reached): PR #9014 makes the Shell tool honor an explicitly configu...: none — all checks I started were completed within budget..
中文说明
已审查。
未审查：build-and-test — Integration Tests (CLI, No Sandbox) was skipped in CI and its suite did not run locally。
未探索到全部深度（达到工具调用预算）：PR #9014 makes the Shell tool honor an explicitly configu...：none — all checks I started were completed within budget.。
— qwen3.8-max via Qwen Code /review (v0.21.10)
qwen-code-ci-bot
left a comment
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
No issues found. LGTM! ✅
Not explored to full depth (tool budget reached): "You are hunting for gaps in the review of PR #9014, which…": none — all checks above completed within budget.; "PR #9014 makes the Shell tool honor a user-configured…": could not execute coreToolScheduler.test.ts to confirm the promote test reaches the new getter read (no node_modules in the review worktree or parent checkout…; "You are hunting for gaps in the review of PR #9014, which…": none — all checks above completed within budget.; "PR #9014 makes the Shell tool honor a user-configured…": did not execute the test suite** — the review worktree has no node_modules (vitest binary missing); verification above is static (test-vs-implementation mutat….
中文说明
未发现问题。LGTM！✅
未探索到全部深度（达到工具调用预算）："You are hunting for gaps in the review of PR #9014, which…"：none — all checks above completed within budget.；"PR #9014 makes the Shell tool honor a user-configured…"：could not execute coreToolScheduler.test.ts to confirm the promote test reaches the new getter read (no node_modules in the review worktree or parent checkout…；"You are hunting for gaps in the review of PR #9014, which…"：none — all checks above completed within budget.；"PR #9014 makes the Shell tool honor a user-configured…"：did not execute the test suite** — the review worktree has no node_modules (vitest binary missing); verification above is static (test-vs-implementation mutat…。
— qwen3.8-max via Qwen Code /review (v0.21.11)
doudouOUC
left a comment
doudouOUC
left a comment
There was a problem hiding this comment.
Reviewed. Suggestions are inline. 1 Suggestion-level finding(s) could not be anchored to a changed line and were dropped; nothing further to act on here.
Not explored to full depth (tool budget reached): PR #9014 makes the Shell tool honor the configured tools...: 无。`.
中文说明
已审查。 建议见行内评论。 1 条建议级发现无法锚定到改动行，已丢弃；此处无需进一步处理。
未探索到全部深度（达到工具调用预算）：PR #9014 makes the Shell tool honor the configured tools...：无。`。
— deepseek-v4-flash via Qwen Code /review (v0.21.10)
…LM#9086) * fix(review): harden the pipeline against four live-run failures Measured on three parallel PR reviews (qwen3.8-max, 2026-08-13, PRs QwenLM#9013/QwenLM#9014/QwenLM#9045) run via `qwen review run`: - run.ts: pin the composed-verdict and report scans to the run's own target. The generic newest-composed scan captured a concurrent run's artifact — two of the three runs republished a neighbour PR's verdict (one reported REQUEST_CHANGES for a review whose own report said Comment). Also keep re-reading while the child runs: a coverage re-check legitimately recomposed a verdict 12 minutes after the first write, and the first-snapshot capture would republish the superseded one. - budget.ts: drop placeholder gaps whose completion word carries a trailing budget adverbial. Three "none — all checks … completed within budget" non-answers reached two posted bodies because the completion idiom required the completion word to end the text. - coverage.ts: label a non-chunk agent by the brief codename found anywhere in its launch prompt. Launchers prepend context lines, so the first-line label gave twelve finders one shared PR-summary sentence, and every budget-gap disclosure rendered as the same truncated PR quote instead of a name. - copy_bundle_assets.js: emit dist/cli.js with a shebang and the execute bit. shellContextEnv blanks a QWEN_CODE_CLI a POSIX shell cannot exec, so every review subcommand issued from a session launched off the bundle silently fell back to the PATH's global install — all three runs executed the machine's auto-updated release instead of the tree they were launched from. * test(review): follow the codename label into compose-review's fixtures The backtick-collapse fixture's first line was itself the brief codename shape, so the new codename extraction labels it `agent security` and the first-line assertion no longer holds. Keep the sanitization intent on a prose-only first line, and pin the codename behaviour — a prepended context line must not reintroduce the shared-PR-quote label — as its own case. * fix(review): classify the run target with the child's own parser Review feedback on the target pin: prNumberFromTarget re-derived PR classification with a narrower regex than parse-args — /pull/<n>/files URLs went unpinned, 0042 pinned pr-0042- while the child writes pr-42-, and docs/pull/42 pinned a file target as a PR — so a completed (and posted) review could be reported as one that produced no verdict. Delegate to parseReviewArgs, whose verdict is what the child names its artifacts from, and pin the divergent shapes as tests. Also gate the bundle's shebang/exec-bit block with a package-assets case (mode asserted off-win32, double-run must not stack shebangs), and document the accepted same-target residual race on composedPatternFor. * fix(review): pin run artifacts by exact target identity, share the identity-line parser Round-2 review feedback, all six findings: - run.ts: replace name-shape pins with the exact composed filename each target class produces (pr-<n> / file basename / the fixed 'local' token, per the skill's --out template). The (?!pr-\d+-) lookahead rejected a file run's own artifact whenever the reviewed file was named pr-<digits>-…, the PR branch's .* wildcard claimed that same artifact, and the pooled null class let concurrent file and no-target runs cross-capture each other's verdicts. Target classification now comes from classifyRunTarget (parse-args' verdict, basename for files). - run.ts: newestArtifactSince returns {path, mtime}, so the capture poll reuses the scan's own stat instead of re-statting the path — the scan-vs-sweep window (and its untestable catch branch) is gone structurally. - lib/agent-identity.ts: one parser for the identity line agent-prompt bakes into every launch, shared by cost-ledger's row labels and coverage's disclosure labels — the two hand-rolled copies could drift, and coverage's copy dropped the (round N) and owned-file suffixes, folding reverse-audit rounds into indistinguishable disclosure lines. cost-ledger still feeds it only the first line (quoted identity lines below must never be credited); coverage scans for the first line-anchored identity line (launchers prepend context lines). - lib/budget.ts: one vocabulary for the budget-idiom family — 'below' joins the completion tail's position words, and the stayed idiom takes the same qualifiers ('stayed inside the tool-call budget'). - run.test.ts: handler-level assertion that the report scan is pinned (a strictly newer neighbour report must not become this run's reportPath), alongside the pattern-level cases for every shape the review probed. * fix(review): round-3 review polish — CRLF identity lines, producer-side no-gap rule, edge-case pins - agent-identity: tolerate a trailing CR (CRLF-recorded prompts fed through \n-splitting callers failed every parse and fell back to first-line prose); scan the launch prompt with one multiline match instead of materializing a line array per agent record; pin the round-over-file precedence with a both-suffixes test. - agent-prompt: state the no-gap rule at the producer — write NO 'Budget gap:' line when nothing was cut short — instead of leaving each agent to improvise a 'none' phrasing the consumer-side placeholder filter must chase forever. - run.ts: strip trailing path separators before taking a file target's basename (a tab-completed 'src/' pinned 'qwen-review--composed.json', which no child artifact carries — fail-closed exit 1 on a completed review); pin file-run reports by their filename slot so a file named 'pr-1234.md' claims its own report; document the two collision classes the basename-keyed pin defines (same-basename files, basenames spelling 'local'/'pr-<n>'). - budget.test: pin the parenthesis-form exception keep case beside the dash form. * fix(review): round-4 polish — one budget vocabulary, named pin expectation, honest prompt claim - budget.ts: spell the budget-position vocabulary once (BUDGET_QUALIFIED / COMPLETION_TAIL) and build PLACEHOLDER_GAP_RE from it — the literal carried three hand-copies that had already drifted twice in two review rounds; the space-separated 'tool call' form is pinned in both branches. - agent-prompt.ts: the no-gap rule now states what actually happens to a 'none' disclosure — at best filtered, and any unrecognized wording is published as a phantom coverage gap — instead of claiming the parser treats it as a gap, which was the negation of the filter shipped beside it. - run.ts: derive the composed pin from composedNameFor and name the expected filename in both the no-verdict prose and the JSON result (expectedComposedName) — a naming drift between the pin and the skill's template was undiagnosable once Step 9 swept the near-miss. - compose-review.ts: publicAgentSubject's provenance note now describes the codename labels coverage prefers, with first-line prose as the fallback. * test(review): guard the no-verdict diagnostic and the pinned capture; align the chunk-role grammar - run.test.ts: assert the no-verdict report names the artifact it waited for, in prose and as expectedComposedName — mutation-verified: dropping the suffix now fails. - run.test.ts: force the neighbour's composed artifact strictly NEWER in the concurrent-run fixture. With it older, an unpinned newest-composed scan landed on the right file anyway and the regression passed; mutation-verified: reverting composedPatternFor to the generic scan now fails the handler test, not only the pattern units. - agent-identity.ts: CHUNK_ROLE_RE takes coverage's CHUNK_RE shape (whitespace-tolerant, case-insensitive) so a hand-edited 'Chunk 3 of 7' cannot resolve as a chunk owner in the posted body and a role agent in the ledger row. * fix(review): keep the bundle's write time across the shebang rewrite; hold the pins to the skill Round-6 review feedback, all five findings: - copy_bundle_assets.js: preserve dist/cli.js's atime/mtime across the shebang rewrite. stampReviewSourceDigest reads that mtime as the build time, so a bumped one certifies a bundle as newer than review sources edited before it and the staleness warning the skill's Step 0 stops on never fires. A full bundle stamps before reaching here, but a standalone run of this script — a flow the gate's own comment contemplates — was exposed. - package-assets.test.js: pin both halves the block owes. The mtime is asserted against a fixture built 60s in the past, and the second run now arrives at mode 0644 so the exec bit must be re-set — demoting the chmod inside the shebang guard previously stayed green. - run-skill-parity.test.ts: new. composedNameFor and reportPatternFor encode the bundled skill's Step 6 --out template and Step 8 report stems, and were pinned only against self-referential literals. This reads the templates out of SKILL.md and renders them per target class, so a skill-side edit fails next to the code that must follow it instead of silently in a later review. - cost-ledger.test.ts: pin the first-line-only invariant — a launch whose prepended context sits above the identity line keeps the transcript's own id, never a label lifted from below. Consolidating both callers on labelFromLaunchPrompt now fails. - agent-identity.test.ts: assert the two entry points genuinely differ on that prompt, so neither caller's policy can be collapsed into the other unnoticed. * fix(review): stop the mtime assertion from pinning libuv's timespec truncation The assertion compared the recorded mtime against the Date handed to utimesSync, so it also pinned libuv's double-seconds → timespec conversion: about half of all millisecond values read back 1 ns low (X - 0.001), and builtAt is a fresh Date.now() - 60_000 every run — a ~50% coin flip that would have landed intermittent reds on unrelated PRs through test:ci. Capture what the filesystem actually recorded after the setup and compare against that; the invariant under test is only whether the shebang rewrite moves the stored time. 10/10 green through the CI entry point, and it keeps its teeth: removing the production fs.utimesSync(cliEntry, atime, mtime) restore fails it 3/3.
yiliang114
left a comment
yiliang114
left a comment
There was a problem hiding this comment.
Approved on 120f56f. Verified end to end:
- Wiring is correct and capture-safe:
getShellOutputThreshold(config)reads the Config at invocation time and is shared by both consumers — the producertruncateToolOutputcall site (shell.ts execute path) and the scheduler-facingmaxOutputCharsgetter — so the two layers cannot diverge and there is no stale capture. - Back-compat verified: with the setting absent,
isTruncateToolOutputThresholdExplicit()is false and Shell keeps its 30K trigger + 4K preview exactly as before. The!= nullflag also treats a runtimenullas unset (regression test in config.test.ts), and both the getter default and the call-site default are pinned by execute-path tests. - Default-injection risk ruled out: the only production Config producer is
packages/cli/src/config/config.ts:2311, which passes the rawsettings.tools?.truncateToolOutputThresholdwith no?? default. The schema'sdefault: 25_000is consumed only by the settings-dialog/config-command UI helper (getEffectiveValue), not the Config build chain — so the Shell 30K default is not silently flipped to 25K for users who never configured the setting. This addresses the open R3-1 / doudouOUC concern; the author's rebuttal checks out. - Invalid values are safe:
getTruncateToolOutputThreshold()maps<= 0toInfinity(matches the documented "Set to -1 to disable"), so 0/negative values disable truncation instead of crashing. Units are characters everywhere, consistent with settings.md and the existing char-based Shell trigger;lines: Infinitykeeps the line cap from undercutting the char budget as before. - Sub-4K thresholds honored:
previewChars: Math.min(4000, outputThreshold)closes the earlier gap where a low explicit threshold was undercut by the fixed 4K preview; truncation-marker and persisted-output-file propagation are covered by the new 10K execute-path test. - All prior review rounds' findings were addressed with tests (author replies cite 3b6e604, 27d9a62, 64ce61e; the final bot round landed on LGTM).
Nits (non-blocking, P3):
- The "producers must not pass a defaulted value" invariant lives only in the config.ts constructor comment. Today there is exactly one producer, but a future Config producer that applies schema defaults would silently move Shell from 30K to 25K. A one-line reminder near cli config.ts:2311 would harden it.
docs/users/configuration/settings.mdreconciliation stays out of scope per the author; note the doc already claims this setting applies to Shell, so runtime now matches doc text — only the Shell-default-30K-vs-25K nuance remains undocumented.
CI: 110/110 check runs completed, zero failures. Test (ubuntu-latest, Node 22.x) and web-shell E2E Smoke passed; macOS/Windows matrix and Integration Tests (CLI, No Sandbox) were skipped under the fork CI-approval gate — small residual risk, covered by the added unit tests.
|
Released in v0.21.12. |
4 participants
What this PR does
This change makes Shell use an explicitly configured
tools.truncateToolOutputThresholdat both output-processing boundaries. When the setting is absent, Shell keeps its existing 30,000-character default; explicit lower, higher, and disabled values take precedence.It also adds regression coverage for configuration provenance and the Shell output path, and updates the design documentation that describes Shell truncation.
Why it's needed
Shell currently hardcodes a 30,000-character threshold, so a 40,000-character result is truncated even when the user configures a 100,000-character threshold. This contradicts the documented setting and unnecessarily replaces model-facing output with a shortened preview and persisted-file reference.
Reviewer Test Plan
How to verify
Configure
tools.truncateToolOutputThresholdas 100,000 andtools.truncateToolOutputLinesas 100,000, restart Qwen Code, and ask Shell to runnode -e "process.stdout.write('x'.repeat(40000))". Confirm that the complete output reaches the model without a truncation marker or persisted-output reference. Then remove the threshold setting and confirm that the same output is still truncated at Shell's existing 30,000-character default. Optionally set the threshold to 10,000 or-1and confirm that Shell respectively truncates earlier or disables character truncation.Evidence (Before & After)
Before: Qwen Code 0.21.9 persisted the 40,000-character result to a
run_shell_command_*.outputfile despite an explicit 100,000-character threshold.After: the current source keeps all 40,000 characters with an explicit 100,000 threshold and creates no artifact. The absent setting still truncates at 30,000; an explicit 10,000 threshold truncates a 20,000-character result; and
-1keeps all 40,000 characters.Tested on
Environment (optional)
macOS arm64, Node.js v24.14.1. The installed 0.21.9 CLI baseline ran under the macOS seatbelt sandbox; the current source was verified through the real Shell subprocess path plus focused unit and scheduler tests.
Risk & Scope
Linked Issues
Closes #8922
中文说明
本 PR 做了什么
本改动让 Shell 在两层输出处理边界都使用用户显式配置的
tools.truncateToolOutputThreshold。未配置时，Shell 继续保持现有的 30,000 字符默认值；显式配置的更低、更高以及禁用值都会优先生效。同时补充了配置来源与 Shell 输出路径的回归测试，并更新了描述 Shell 截断行为的设计说明。
为什么需要
Shell 当前硬编码了 30,000 字符阈值，因此即使用户配置为 100,000，40,000 字符的结果仍会被截断。这与设置文档不符，也会不必要地将模型可见输出替换成缩短预览和持久化文件引用。
Reviewer 测试计划
如何验证
将
tools.truncateToolOutputThreshold和tools.truncateToolOutputLines都配置为 100,000，重启 Qwen Code，然后让 Shell 执行node -e "process.stdout.write('x'.repeat(40000))"。确认完整输出进入模型上下文，且没有截断标记或持久化输出引用。随后移除阈值配置，确认相同输出仍按 Shell 现有的 30,000 字符默认值截断。也可以将阈值设为 10,000 或-1，确认 Shell 分别更早截断或禁用字符截断。证据（修复前后）
修复前：Qwen Code 0.21.9 在显式配置 100,000 字符阈值时，仍会将 40,000 字符结果持久化为
run_shell_command_*.output文件。修复后：当前源码在显式配置 100,000 时保留全部 40,000 字符且不创建 artifact。未配置时仍按 30,000 截断；显式配置 10,000 时会截断 20,000 字符结果；配置
-1时会保留全部 40,000 字符。测试平台
环境（可选）
macOS arm64，Node.js v24.14.1。已安装的 0.21.9 CLI 基线在 macOS seatbelt sandbox 下运行；当前源码通过真实 Shell 子进程路径以及定向单测和 scheduler 测试完成验证。
风险与范围
关联 Issue
Closes #8922