batch-plan-execute（批量计划执行）

Batch Plan Execute

Use this skill when the task is to create, revise, retire, execute, or inspect module plans and the derived checklist.md based on a requirement source and the current plans/ directory state.

This skill has two explicit modes:

• plan: generate or revise implementation-ready plans and refresh checklist.md
• execute: implement already-reviewed plans

This skill can:

• accept requirement text from chat, a requirement file such as requirements.md, or an existing plan file or plans/ directory
• reconstruct the current planning state from the latest requirement source, existing plan files, and a hidden state file
• detect mixed changes per module: new requirements, changed requirements, removed requirements, and review notes
• build dependency-aware module or workstream plans that reflect serial and parallel work safely
• generate a consolidated checklist.md from the latest requirement source plus authoritative review notes
• execute approved plans with subagents only after an explicit execution command from the user

Inputs And Resolution

Primary Input Resolution

Choose the primary input in this order:

1. an explicit plan file path or plans/ directory path mentioned by the user
2. an explicit requirement file path mentioned by the user
3. a file or document already attached in the task context
4. the requirement text written directly in the chat

Use these rules:

• If a plan file path is provided, treat that plan lineage as the primary starting point and inspect the sibling plans/ directory.
• If a plans/ directory path is provided, inspect only plan files and the state file inside that directory.
• If a requirement file path is provided, read it directly and treat it as the latest requirement source.
• If the input is raw chat text, treat the full user text as the latest requirement source.
• If both a requirement source and a plans/ directory are available, always use both.
• Fail immediately if any referenced file or directory does not exist or is not readable.

Output Language Resolution

Use these rules:

• Match all user-visible output language to the dominant language of the latest requirement source.
• If no requirement source exists, match the dominant language of the user's current-turn input.
• If the user explicitly asks for translation or a specific output language, follow that instead.

Mode Routing

Default to plan mode.

Switch to execute mode only when the user explicitly issues an execution command in the current turn.

Examples that DO resolve to execute mode:

• 开始执行
• 按这个方案实现
• 去改代码
• 直接落地
• implement now
• apply the plan

Examples that DO NOT change mode from the default plan behavior:

• LGTM
• OK
• 继续
• 按这个方向
• review 过了
• any approval that confirms the plan but does not explicitly ask to start implementation

Use these rules:

• Review approval is not execution approval.
• If the user is ambiguous, stay in plan mode and either refine the execution plan or ask for a clearer execution command.
• Never infer execution intent from momentum, optimism, or lack of objections.

Shared Semantic Model

Requirement Preprocessing

Before extracting requirements, computing requirement-derived hashes, deriving checklist items, or interpreting review notes, preprocess the latest requirement source in this order:

1. remove HTML comments outside fenced code blocks
2. ignore Markdown structure inside fenced code blocks
3. exclude document-frontmatter --- delimiters from requirement segmentation
4. detect standalone --- lines in the remaining requirement body as requirement-item dividers only when they are outside fenced code blocks, outside removed comments, and surrounded by non-empty requirement content

Use these rules:

• Treat commented requirement content as nonexistent.
• HTML comments inside fenced code blocks are not review notes.
• Review comments are only meaningful inside plans/*.md; requirement sources use the same comment-boundary semantics for preprocessing only.
• Empty divider segments, comment-only divider segments, and formatting-only divider changes must not create semantic requirement changes.

Requirement Item Identity And Segmentation

Use these rules:

• Treat each standalone ----delimited body segment as a distinct requirement item candidate before module mapping.
• Treat each standalone ----delimited body segment as a new planning batch boundary, not merely as an extraction aid.
• A new standalone ----delimited requirement segment must produce a new requirement item identity and a new plan lineage, even when its implementation area overlaps with an earlier module.
• Do not merge a requirement item extracted from a later --- segment back into an earlier plan lineage.
• Requirement item state must preserve extraction-time identity, including item_id, source_kind, source_order, source_excerpt_hash, body_hash_without_comments, mapped_module_slug, and planning_batch_id.
• Meta sections such as background, goals, scope, non-goals, assumptions, rollout, acceptance criteria, and non-functional requirements are global constraints unless they clearly describe implementation-bearing work.

Module Lineage And Ownership

Use these rules:

• Module boundaries should follow implementation ownership, shared code surface, migration boundaries, deployment boundaries, and owner-level deliverables rather than raw heading order.
• Plan slugs should follow implementation ownership, not requirement heading text, when the two differ.
• Merge requirement slices that land on the same implementation path, shared code surface, or owner-level deliverable.
• Split only when the source clearly describes independent deliverables or strict prerequisite sequencing.
• Do not create overlapping modules that both own the same shared change.
• If a shared change is needed by multiple modules, assign one clear owner module lineage.
• Resolve latest lineage artifacts by highest rev, otherwise latest base file.
• Prefer slug matches first.
• Allow rename matching only for safe one-to-one matches on body_hash_without_heading.
• If rename matching is ambiguous, do not guess.

Dependency Layering And Parallelism

After extracting candidate modules, derive an explicit dependency graph before classification or subagent spawning.

Use these rules:

• Classify each module as foundation, feature, or follow-on.
• Record depends_on, blocks, parallelizable_with, and shared_change_owner.
• Prefer parallel modules only when they do not require the same shared code ownership in the same phase.
• Prefer serial sequencing when one module changes contracts, schemas, migrations, shared utilities, or infrastructure that another module consumes.
• If the implementation is A then B, model that directly instead of forcing both into one peer list.
• Do not infer fake parallelism. Sequence conservatively if safe ordering is unclear.

Artifact Model And File Contracts

Plan File Lineage

Write plan files into a plans/ directory:

• If the source is a requirement file, use a plans/ folder next to that file.
• If the source is a plan file or plans/ directory, keep writing into that same plans/ directory.
• If the source is raw chat text, use <cwd>/plans/.

Create the plans/ directory if it does not exist.

Base plan files use this format:

• <module-slug>.md

Revision plan files use this format:

• <module-slug>.rev-<n>.md

Use these naming rules:

• new-plan may create a base file if the module has no prior lineage.
• revise-plan and obsolete-plan must always write a new rev file.
• no-op must not write a new plan file.
• Do not overwrite an existing file silently.
• A new standalone ----delimited requirement segment must create a new base file and a new plan lineage.
• Do not route a later ----delimited requirement segment into an earlier lineage, even if it touches the same implementation area or extends the same feature family.
• Revisions apply only within the lineage created from that specific requirement segment after its base file exists.
• Slugs derived from ----delimited requirement segments must still describe the implementation owner-level deliverable, not the raw segment order or divider label.

Checklist Contract

The derived checklist file must be:

• checklist.md

Use these rules:

• If the latest requirement source is a readable file, write checklist.md next to that requirement file instead of inside plans/.
• If no readable requirement file exists, fall back to writing checklist.md inside the active plans/ directory.
• Always overwrite the resolved checklist.md path in place when it can be refreshed safely.
• checklist.md is a derived acceptance snapshot based on the latest requirement source and authoritative review notes.
• It exists to help humans verify whether the requested outcomes and acceptance expectations have been satisfied.
• It is not a plan lineage artifact, not an execution target, and not an implementation task list.
• Review notes may refine or correct checklist content only when they change the authoritative requirement or acceptance intent.
• If an earlier checklist.md exists, treat it only as an optional source of prior checkbox state for safely matched unchanged items, never as the source of truth for checklist content.

State File Contract

The hidden state file must be:

• .state.json

Maintain plans/.state.json as the planning baseline.

The state file must contain at least:

• requirement_source
• requirement_hash
• run_at
• requirement_items
• modules

Each module state object must contain at least:

• slug
• title
• status
• source_excerpt_hash
• body_hash_without_heading
• requirement_item_ids
• latest_plan_file
• latest_rev
• last_action

Status values must be explicit:

• active
• obsolete

Use these rules:

• The state file describes planning lineage only.
• Do not treat the state file as permission to execute.
• If the state file is missing, bootstrap from the latest requirement source and the existing plan files instead of failing.
• Bootstrap must reconstruct both requirement items and module mappings using the same shared semantic model as normal plan mode.
• If the state file exists but is malformed and cannot be safely reconstructed from visible artifacts, fail.

Plan Mode

Use docs/plan.md as the detailed plan-writing and plan-refresh contract.

Plan mode is responsible for:

• extracting implementation-bearing modules from the latest requirement source under the shared semantic model
• classifying each module into exactly one action: new-plan, revise-plan, obsolete-plan, or no-op
• producing implementation-ready module plans
• refreshing checklist.md
• refreshing plans/.state.json after all affected modules are processed

Use these rules:

• Treat authoritative review notes from the latest plan lineage as input constraints for plan revisions.
• Merge requirement changes and review notes into one revised module plan when both affect the same lineage.
• Generate or refresh checklist.md during every successful plan run.
• Keep detailed module extraction, checklist refresh, subagent planning, and plan assembly behavior in docs/plan.md rather than duplicating it here.

Execute Mode

Use docs/execute.md as the detailed execution contract.

Execute mode is responsible for:

• resolving the target plan file or plans/ directory
• selecting the latest reviewed artifact for each chosen module lineage
• dispatching execution directly from approved plan boundaries and dependency metadata
• validating the implemented change before reporting completion

Use these rules:

• Execute only after an explicit execution command in the current turn.
• Execute the latest reviewed artifact for each selected module lineage.
• Fail immediately if the latest target artifact still contains unresolved review comments.
• Treat approved plans as the execution boundary source of truth. Do not re-scope modules, re-derive write ownership, or re-plan dependency layering during execute mode.
• Keep target resolution, preflight, worker dispatch, integration, verification, and completion details in docs/execute.md rather than duplicating them here.

Failure Conditions

Stop and report the problem instead of generating weak output when:

• the requirement file, plan file, or plans/ directory is missing or unreadable
• module extraction fails for a required requirement source
• a plan would be mostly speculation with no meaningful codebase tie-in
• multiple latest lineage candidates exist for one module and the ordering cannot be resolved safely
• rename matching is ambiguous and cannot be reduced to a safe one-to-one mapping
• shared changes or dependency ownership cannot be assigned to one clear module or workstream
• the dependency graph contains cycles that cannot be justified as a single merged module
• the state file exists but is malformed and cannot be safely reconstructed from visible artifacts

Do not emit empty plan files, summary-only files, or a single aggregate plan when the task requires per-module outputs.