designcraft-use

Page layout and publishing

Candidate implementation, actual host, native installation, and creative acceptance not yet completed. macOS arm64, Python 3.11+ only.

Routing scope

This is the sole default entry point when the user raises a DesignCraft new-document, page-design, full-layout, or publishing goal. Even if the user only mentions "two pages," "editable document," or creating a new file, or has not yet submitted an outline, as long as content/layout planning is needed, first clarify content, pages, assets, scope of changes, and deliverables at this entry point.designcraft-cli-document Handles only single lifecycle operations on existing projects, and is not responsible for creating blank projects or new documents. Local layout adjustments are handed to designcraft-cli-layout（Install: npx skills add full-aigc-skills/designcraft-skills --skill designcraft-cli-layout), and explicit exports of existing projects are handed to designcraft-cli-export（Install: npx skills add full-aigc-skills/designcraft-skills --skill designcraft-cli-export). When the user explicitly asks only about native commands or installation diagnostics, hand off to the corresponding atomic skill; unrelated requests do not trigger this skill. Execution may install runtimes, modify projects, and write output artifacts; proceed only within authorized paths. For unknown execution results, first verify the persistent project and receipts, and do not automatically replay writes. Completion is determined by saving and reopening the project, verifying exports, and reviewing page by page; command success does not mean acceptance is complete.

When to Use

Use this entry point when the user wants to create a new document, or produce or revise pages starting from goals/outlines/assets; when input is missing, first ask for the specific missing items. Even if the user only says "create a blank document," this entry point still receives it and clarifies the goal. Only when the request requires just one native lifecycle action on an existing project, standalone installation diagnostics, command queries, local operations on existing objects, or export of an existing project, hand it off to the corresponding atomic skill.

Out of scope and safety boundaries

• Questions unrelated to DesignCraft do not trigger this skill; do not treat planning suggestions as user authorization.
• Do not infer fixed CLI parameters from examples or research source code; before execution, use the native directory of the current fixed version.
• Scripts read and write local files with the current user's permissions. Registered inputs and digests can only detect partial drift and cannot serve as a sandbox, rollback, or permission isolation.
• When dependency installation authorization is missing, perform offline checks only; do not automatically download, install, log in, or upgrade.

Workflow

Step 1: Clarify scope

Record assets, page count/dimensions, target format, output location, originals to preserve, and allowed scope of modification; for output to the same path, first verify overwrite authorization.

Step 2: Check prerequisites

Confirm host loading, fixed runtime status, and shared receipt contract. When installation authorization is missing, remain not ready, and do not substitute offline directories or research source code for real capabilities.

Step 3: Query and plan

Use this skill's entry point to read the current CLI command directory and parameters, then form a single-session plan; when existing projects are involved, register the originals and use a working copy.

Step 4: Execute and retain evidence

Execute the plan only within the authorized scope, recording receipts, input/resource digests, and newly created artifacts; on timeout, interruption, or incomplete identity, retain UNKNOWN and do not replay writes.

Step 5: Independent acceptance

Save and reopen the project in a new session, verify page count, text, linked assets, and exported files, then review page by page. If any item is unverified, continue to mark it as pending acceptance.

Rules (execution constraints)

• Native directory queries can be offline; actual execution must use the real directory of the current pinned CLI.
• Runtime, skill snapshot, host discovery, model routing, native commands, project reopen, and visual review are independent gates.
• A save command or export command exiting with zero only proves the process result, not that content or visual quality passed.

Gotchas (common misconceptions)

1. SKILL_DIR must come from the actual load path: do not guess sibling skill directories or development workspace paths.
2. Offline catalog cannot be used for execution: stop writing when the real CLI or real command parameters are missing.
3. UNKNOWN is not an ordinary failure: first read the receipt and verify the existing project; do not rerun a plan that may have already produced side effects.
4. A digest does not imply isolation or rollback: unregistered inputs, paths hidden in parameters, and their side effects are not protected by the digest.
5. A pinned version does not mean the host has accepted it: plugin discovery, model selection, and real native command invocation must each be evidenced separately.

First use

Set SKILL_DIR to the absolute directory where the SKILL.md actually loaded by the host resides. The skill can be installed independently into the project .agents/skills, the user skill directory, or the plugin cache; do not guess sibling directories, and do not modify skill files. When the user request has authorized the necessary dependencies, the public cli.py automatically verifies and installs the pinned native version into the user data directory; do not use sudo, and do not modify PATH.

: "${SKILL_DIR:?设置为本技能真实加载目录}"
python3 -I -B "$SKILL_DIR/scripts/cli.py" -- commands

For installation diagnostics and isolated runtime parameters, see the required pre-execution reading.

Current scenario

First record the user's materials, goals, output location, dimensions/page count/format, and modification boundaries. Query the real parameters from the native directory and execute after confirming the command; do not fabricate commands, and do not treat new interfaces in local source code as already supported by the pinned artifact. The material paths and output locations in the examples below must be replaced with the actual authorized paths for the current task.

python3 -I -B "$SKILL_DIR/scripts/cli.py" -- commands

Deliver the native .designcraft project, a collected materials and font license list, page-by-page previews, and applicable PDF/IDML/EPUB. Reopen the native project to verify page count, text, and linked materials; record interchange format losses separately.

CLI parameters are passed as argv, without executing a shell. Native commands may have file write and network behavior; use them only within the authorized scope of the task. If execution fails, first verify the project and output; do not replay writes when the result is unknown. Protect existing files before outputting to the same path. A command query or zero exit does not mean delivery passed: inspect the actual files, save and reopen, and the content and post-modification results, and separately record the creative review. Runtime and skill installation, host skill discovery, and model natural language selection are gates accepted separately.

Scenario examples

See the successful path for multi-page layout orchestration of this skill and the refusal and recovery boundaries for irrelevant requests/unknown results in Scenario examples、From outline to editable project、Local revision of an existing projectandUNKNOWN recovery. The examples only illustrate the flow and do not prove that the pinned native CLI has been accepted.

For the input registration scope, working copy strategy, pre- and post-execution digests, and receipt field descriptions, see this skill's Execution and receipt reference. Digests are only for drift detection and do not constitute proof of a sandbox or rollback.

Required pre-execution reading

Before executing a command plan, checkpoint recovery, or delivery verification, you must read this skill's execution and acceptance details, and operate according to its requirements for input preservation, UNKNOWN, receipts, and format verification. Summaries can only detect changes; they cannot isolate or roll back writes; unknown results must not be automatically replayed.