Skip to content

feat(sheets): eval round 2 — cut agent error rate and --help lookups - #1911

Merged
xiongyuanwen-byted merged 4 commits into
feat/lark-sheets-developfrom
feat/sheets-round2-optimization
Jul 16, 2026
Merged

feat(sheets): eval round 2 — cut agent error rate and --help lookups#1911
xiongyuanwen-byted merged 4 commits into
feat/lark-sheets-developfrom
feat/sheets-round2-optimization

Conversation

@xiongyuanwen-byted

@xiongyuanwen-byted xiongyuanwen-byted commented Jul 16, 2026

Copy link
Copy Markdown
Collaborator

Background

Round 2 of eval-driven sheets optimization, rebased onto the latest feat/lark-sheets-develop (8897196d).

Changes

  • feat(sheets): cut agent error rate and --help lookups (eval round 2) — targets the top failure modes from round 2 evals, reducing agent error rate and the number of --help lookups.
  • chore(sheets): sync skill docs and flag data from sheet-skill-spec — syncs skill docs and flag data from sheet-skill-spec.

Notes

  • During rebase, the "import mislabeled .xls workbooks by sniffing content" fix already existed on the target branch (identical patch-id), so it was auto-skipped — no duplicate.
  • The target branch was force-rewritten and advanced in the meantime; the two new commits were cleanly replayed onto the new tip via --onto with no conflicts. One hunk touching the --type description in lark-sheets-workbook.md was auto-dropped because upstream already has the same end state — no content lost.

基于 2026-07-09 三轮评测(35 题、100 次错误、90 次 help/schema 探查)逐条归因:

- 直觉 flag 两级处置(flag_ergonomics):无歧义别名静默改写零来回
  (--file→--csv、--name→--title、--cols/--rows→--range、
  --source/--target→--source-range/--target-range),有歧义的挂
  curated 处方一步给正确写法(sheet-copy 新表名、dim-insert 无
  --dimension、dim-freeze --frozen-rows、cells-set-style --bold、
  table-put --start-cell/--sheet-name);评测 18 次 unknown flag
  did-you-mean 仅命中 4 次
- 下划线 flag 形态全域归一(--sheet_name→--sheet-name),枚举别名
  补 MERGE_ALL 族与 word-wrap true/false
- --cells 单 cell 对象自动包裹为 [[cell]](键集⊆cell schema 才改写);
  border_styles.all 展开四边、cell_styles 误嵌 border_styles 拦截
  (覆盖 standalone 与 batch 嵌套,评测中此类错误须到服务端才拦)
- table-put:--sheets 解码错误内联骨架示例;--styles 一次聚合报告全部
  问题且 type/size 错误内联完整合法 op(评测 V2U011/V2U032 各连环
  踩 4 次)
- batch 服务端「N succeeded, M failed」双层转义 JSON 解包为单行失败
  清单(评测聚合脚本都抓不到原因的 8 次全部来自此处)
- chart:颜色缺 # 前缀自动补全;--print-schema 支持点分路径切片
  (properties.snapshot.plotArea.axes);+chart-create --print-example
  输出 8 类图表最小可用模板(25 次全量 schema 翻页的根源)
- 确认门 hint 携带完整重试命令(含 stdin/超长两个省略保护);
  评测 9 次 requires confirmation 全部靠模型自行补 --yes 自愈
- 高频命令 --help 补可抄示例 Tips(dim-insert/dim-freeze/cols-resize/
  cells-set/cells-set-style/cells-batch-set-style/sheet-copy/table-put)
同步 sheet-skill-spec feat/sheets-round2-optimization 的生成产物:
SKILL.md 瘦身 14%(37.2k→31.8k 字符)+ 高频命令签名速查 +
high-risk-write --yes 标注 + print-example/schema 切片指引。
@coderabbitai

coderabbitai Bot commented Jul 16, 2026

Copy link
Copy Markdown

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 5e65b5a4-a931-4d83-a766-b55bd701ed83

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

  • 🔍 Trigger review
✨ Finishing Touches
🧪 Generate unit tests (beta)
  • Create PR with unit tests
  • Commit unit tests in branch feat/sheets-round2-optimization

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

Comment @coderabbitai help to get the list of available commands.

@github-actions github-actions Bot added domain/ccm PR touches the ccm domain size/XL Architecture-level or global-impact change labels Jul 16, 2026
@xiongyuanwen-byted xiongyuanwen-byted changed the title feat(sheets): eval round 2 优化——降低 agent 报错率与 --help 查询 feat(sheets): eval round 2 — cut agent error rate and --help lookups Jul 16, 2026
@xiongyuanwen-byted
xiongyuanwen-byted force-pushed the feat/sheets-round2-optimization branch from e8c4511 to d28d6c9 Compare July 16, 2026 06:59
WithHint("add --yes to confirm")
err := errs.NewConfirmationRequiredError(errs.RiskHighRiskWrite, action,
"%s requires confirmation", action)
if retry := retryCommandWithYes(os.Args); retry != "" {

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

[P1] Avoid echoing the complete argv in confirmation errors

RequireConfirmation embeds the full os.Args in a typed error hint that is JSON-encoded to stderr. High-risk commands accept inline sensitive payloads (for example, apps +db-execute --sql and the generic service --data flag), so any invocation below the 300-byte cutoff can copy SQL, request bodies, or secrets into logs and agent transcripts.

Please keep the fixed add --yes to confirm hint and let callers reuse their original argv array with --yes appended, instead of echoing the full command.

…g path

+cells-set-style / +cells-batch-set-style parse borders via
borderStylesFromFlag, which — unlike the typed --cells and --styles paths —
never expanded the "all" side shorthand, so {"all":…} reached the backend to
be rejected. Expand it CLI-side here too for parity.
retryCommandWithYes only recognized a bare "-" token, so a stdin value
bundled onto a flag as --flag=- slipped through and produced a retry line a
bare re-run could not reproduce. Suppress the line for the bundled form too.
@xiongyuanwen-byted
xiongyuanwen-byted merged commit 482e835 into feat/lark-sheets-develop Jul 16, 2026
4 checks passed
@xiongyuanwen-byted
xiongyuanwen-byted deleted the feat/sheets-round2-optimization branch July 16, 2026 07:31
xiongyuanwen-byted added a commit that referenced this pull request Jul 20, 2026
## Background

Round 2 of eval-driven sheets optimization, rebased onto the latest `feat/lark-sheets-develop` (`8897196d`).

## Changes

- **feat(sheets): cut agent error rate and --help lookups (eval round 2)** — targets the top failure modes from round 2 evals, reducing agent error rate and the number of `--help` lookups.
- **chore(sheets): sync skill docs and flag data from sheet-skill-spec** — syncs skill docs and flag data from sheet-skill-spec.

## Notes

- During rebase, the "import mislabeled .xls workbooks by sniffing content" fix already existed on the target branch (identical patch-id), so it was auto-skipped — no duplicate.
- The target branch was force-rewritten and advanced in the meantime; the two new commits were cleanly replayed onto the new tip via `--onto` with no conflicts. One hunk touching the `--type` description in `lark-sheets-workbook.md` was auto-dropped because upstream already has the same end state — no content lost.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

domain/ccm PR touches the ccm domain size/XL Architecture-level or global-impact change

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants