From 5fb12cd6a5a4b6cdedd42c7c97a2f2806e5c669d Mon Sep 17 00:00:00 2001 From: mes123456 Date: Fri, 28 Aug 2026 20:18:57 +0800 Subject: [PATCH] Document efficient continuous execution prompt --- nextTask.md | 79 ++++++++++++++++++++--------------------------------- 1 file changed, 29 insertions(+), 50 deletions(-) diff --git a/nextTask.md b/nextTask.md index b5d279011..b5648682a 100644 --- a/nextTask.md +++ b/nextTask.md @@ -18,57 +18,36 @@ and (when authorized) repository push: ```text 按 nextTask.md 接续执行。 -Treat this as an execution command, not a request for a plan or status report. -Enter CONTINUOUS mode and repeat one bounded task iteration at a time until a -stop condition is reached: +这是执行命令，不是计划或状态咨询。进入 CONTINUOUS 模式，每次只完成一个有边界的任务，成功交接后立即从新 parent manifest 继续，直到全部任务完成、队列闭合或遇到真实停止条件。 -1. Read `docs/EXECUTION_QUEUE.md`; run `node tools/web/print-task-context.mjs` - and `node tools/web/check-task-context.mjs` before editing. -2. Read only the current task card, parent manifest/status, and the focused - files named by the task context. Never open the full - `next-task-plan.json`, all `docs/status` logs, or historical plans. -3. Complete exactly the current task using the commands in its task card. Legacy - `M*-GAP-*` tasks are coverage-only; capability `WBV2-*` tasks own production - behavior and may use shared desktop/WASM/Chromium runners. Do not run a - desktop fixture or comparator unless the current card requires it. - Chromium is the only permitted browser runtime. -4. Keep long output in `tests/golden//command.log`; report only exit - codes, short tails, artifact paths, and SHA-256 values. -5. After every exit criterion passes, run - `node tools/web/check-task-context.mjs --task --write`, write the - report/status/manifest/task-context handoff, and verify all artifact hashes. -6. Run every post-handoff gate: - - ```bash - node tools/web/check-execution-control-plane.mjs - node tools/web/check-blender-next-task-plan.mjs - node tools/web/check-task-index.mjs - npm --prefix web run test:task-context - npm --prefix web run test:context-governance - node tools/web/check-context-governance.mjs - git diff --check - ``` - -7. Only after all gates pass, use repository generators for the task - index/catalog/card and advance `docs/EXECUTION_QUEUE.md` to the newly - completed manifest's `nextTask`. Do not manually edit task-index offsets, - catalog lines, parent evidence, or a guessed task number. -8. If repository saving has been authorized, commit the successful handoff and - push the current branch to its existing `origin`; use the credential helper - or an interactive prompt, never write credentials into files, URLs, logs, or - prompts. Verify `HEAD` and `origin/` match after the push. -9. After a successful handoff, discard the previous task transcript and begin - the next iteration from a fresh `print-task-context` package. Do not wait for - another user message and carry forward only the new parent manifest/status. - -On any failed command, missing focused entry, unavailable environment, budget -overflow, hash drift, governance failure, cancellation, or inconsistent queue, -keep the task `in_progress`/`blocked`, do not advance `nextTask`, and stop with -the exact blocker and evidence paths. Do not fake completion or retry an -unknown state-changing command. For stream-disconnect, error-decoding, or -remote-compact failures, use the recovery procedure below in a new request. -End with a compact summary of completed tasks, states, nextTask, command exit -codes, artifact hashes, push status, and any remaining blocker. +1. 先读取 `AGENTS.md` 和 `docs/EXECUTION_QUEUE.md`，然后运行： + `node tools/web/print-task-context.mjs` + `node tools/web/check-task-context.mjs` +2. 当前任务必须等于队列中 parent manifest 的 `nextTask`。只读取当前任务卡、parent manifest/status 和 task-context 列出的最小输入；禁止默认读取完整历史计划、全部 `docs/status` 或完整 6900 条目详情。 +3. 每次只执行当前任务卡规定的一个行为或一个独立证据变化。`M*-GAP-*` 只作回归覆盖，`WBV2-*` 才负责生产能力；只使用 Chromium，禁止 Firefox/WebKit。 +4. 绝不重复工作： + - 先比较 source hash、behavior hash、dedupe key 和已有 manifest；相同证据直接复用； + - 已有相同行为的 Blender/WASM/Chromium 运行不得再次执行； + - 6900 条目按 capability family 和唯一行为批次推进，不逐条重复验证； + - 已完成条目只作回归引用；代表性回执不能升级为全家族完成。 +5. 新行为必须有真实源码或 WebEngine Main 路径、Chromium 生产测试、malformed/unsupported 失败边界、事务或 undo/redo、save/reopen，以及第二次相同运行的确定性证据。 +6. 只有一对一且 hash-bound 的完整证据才能写入 completion overlay。原始 P0 inventory 和历史 provenance 保持只读；`effective completed/pending` 只能由 overlay 计算，不能篡改原始 6900/6618 基线。 +7. 长输出写入 `tests/golden//command.log`，交互只报告退出码、短尾部、构件路径和 SHA-256。 +8. 任务成功后运行： + `node tools/web/check-task-context.mjs --task --write` + `node tools/web/check-context-governance.mjs` + `node tools/web/check-execution-control-plane.mjs` + `node tools/web/check-v2-task-index.mjs` + `node tools/web/check-v2-handoff.mjs` + `npm --prefix web run test:context-governance` + `npm --prefix web run test:task-context` + `npm --prefix web run typecheck` + `npm --prefix web run lint` + `git diff --check` +9. 只有所有门禁、构件 hash 和 manifest 引用通过后，才使用仓库生成器更新 index/catalog/card、提交并推送现有 `origin/main`。凭据只能交互输入，不能写入文件、URL、日志或提示词；推送后核对 `HEAD == origin/main`。 +10. 成功交接后丢弃旧任务上下文，重新运行 `print-task-context`，只携带新的 parent manifest/status，不能等待下一条用户消息。 +11. 任一命令失败、输入缺失、预算超限、hash 漂移、治理失败、队列不一致、环境不可用或状态不明时，保持 `in_progress`/`blocked`，不得推进 `nextTask`，不得伪造完成或重试未知的状态改变命令，并报告精确阻塞证据。 +12. 每轮结束报告：已完成任务、当前任务、`nextTask`、验证退出码、effective inventory、原始 inventory、构件 hash、提交/推送状态和剩余阻塞。 ``` ## Resume Command