菜单
Publish markdown or HTML content to Confluence pages through the Confluence REST API. Use when the user asks to publish, create, update, sync, or push architecture content, documents, ADRs, designs, or generated toolkit artifacts to Confluence.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Write target architecture documents that connect business capabilities to strategy, target architecture, applications, data, integration, technology direction, and epics to build. Use when the user asks to create, draft, or update a target architecture document, solution architecture document, domain architecture document, capability-aligned architecture document, or TOGAF-style Preliminary and Phase A-E architecture input.
Create Draw.io diagrams from reusable architecture diagram templates and store editable `.drawio` files beside architecture documents. Use when the user asks for Draw.io, diagrams.net, editable architecture diagrams, or when an architecture workflow needs Draw.io versions of target architecture, solution architecture, data architecture, capability context, application component, conceptual data model, integration design, or integration flow diagrams.
Write solution architecture design documents from an existing capability overview, covering implementation design, applications, data model, integrations, diagrams, NFRs, and technical design choices. Use when the user asks to create, draft, or update a solution architecture design, capability architecture document, technical capability design, implementation architecture, or detailed technical design for a business capability.
Create stakeholder-ready problem-to-solution briefs that frame a problem, compare solution options with pros and cons, and recommend a way forward. Use when the user asks to address a problem, brainstorm possible solutions, evaluate alternatives, prepare a senior stakeholder decision paper, or write a recommendation brief.
Create data architecture design documents for a specific canonical data object, including data flow diagrams, integration traceability, ownership, lifecycle, quality, privacy, and governance. Use when the user asks to design, document, review, or update the architecture of a data object and link it to Target Architecture Phase C.
Create integration design documents for APIs, events, files, batches, messaging, orchestration, and system-to-system data flows. Use when the user asks to design, document, review, or update an integration, interface contract, integration flow, source-to-target exchange, API/event/file transfer, or operational integration behavior.
| name | publish-to-confluence |
| description | Publish markdown or HTML content to Confluence pages through the Confluence REST API. Use when the user asks to publish, create, update, sync, or push architecture content, documents, ADRs, designs, or generated toolkit artifacts to Confluence. |
Publish a markdown or HTML file to Confluence using the helper script:
python3 skills/publish-to-confluence/scripts/publish-to-confluence.py path/to/document.md \
--space-key ARCH
When this toolkit is used as a submodule in a private lab repository, run the script from the private lab root:
python3 toolkit/skills/publish-to-confluence/scripts/publish-to-confluence.py outputs/design.md
CONFLUENCE_BASE_URL, such as https://example.atlassian.net/wiki.CONFLUENCE_EMAIL and CONFLUENCE_API_TOKEN for Confluence Cloud basic authentication.CONFLUENCE_SPACE_KEY for the default Confluence publishing space..md, .html, or .htm file..Overview.Never commit API tokens, credentials, private Confluence URLs, customer names, or real company content to this public toolkit repository.
The helper loads Confluence settings from .env in the current working directory. When values are missing, it asks the user for them and stores the missing values in that .env file with owner-only file permissions.
When this toolkit is mounted as toolkit/ in a private lab repo, always run the command from the private lab root so credentials are stored in <private-lab-root>/.env, not in the public toolkit submodule.
The expected .env keys are:
CONFLUENCE_BASE_URL="https://example.atlassian.net/wiki"
CONFLUENCE_EMAIL="name@example.com"
CONFLUENCE_API_TOKEN="..."
CONFLUENCE_SPACE_KEY="ARCH"
Use --env-file <path> only when the private repo keeps local secrets somewhere other than <private-lab-root>/.env.
Use --space-key <key> when publishing to a different space for one run. If .env does not yet contain CONFLUENCE_SPACE_KEY, the script stores the provided --space-key value there.
--env-file pointing to the private repo's local .env.# L2 - Promotions becomes L2 - Promotions. The first line is not included in the published Confluence body.Confluence Link, use the page ID from that link as the target page to update.Confluence Link. After the user provides it, update the source markdown first, then publish to that page.Confluence Link..drawio files for embedded SVG diagrams, and local Markdown links to .drawio or .svg diagrams. Only ask about committing when Git reports open changes for those files. If any of those files have changes, explicitly ask the user whether to commit those publish-related files before publishing. If the user chooses yes, ask for a commit message and create the Git commit. If the user chooses no, skip the Git commit and continue publishing.--dry-run first when the target page, parent page, or conversion output is uncertain.--overview-title <title> when the overview page is not named Overview.The script uses --title when creating a new page. Otherwise, it derives the new page title from the first non-empty line of the source document. When updating an existing page from a Confluence Link, the script preserves the existing Confluence page title and does not rename it. For markdown publishing, the first non-empty line is removed from the Confluence body so the page title does not appear twice.
When a top metadata table includes a Confluence Link row, the script extracts the page ID from URLs such as https://example.atlassian.net/wiki/spaces/ARCH/pages/677838849/L2+-+Promotions and updates that page directly. In that example, the page ID is 677838849.
For markdown publishing, a leading Field / Value metadata table is also removed from the Confluence body. The table can stay in the source file for toolkit tracking without being shown at the top of the published page.
For markdown publishing, links to local .md files are rewritten before conversion. The script resolves each local markdown link relative to the source file, reads the linked file's top Field / Value metadata table, and replaces the markdown file target with that file's Confluence Link. If the linked file does not exist or has no usable top-table Confluence Link, the published body keeps the link text as plain text and does not create a Confluence link.
For markdown publishing, local image references such as  or  are uploaded to the target Confluence page as attachments. The published page body is rewritten to use Confluence attachment image macros so the images render from the page attachments. If an attachment with the same filename already exists on the page, the script uploads a new attachment version. Missing local image files stop the publish.
Markdown or HTML same-page anchors are rewritten to Confluence storage format. Links such as [ARCH001](#arch001) become Confluence anchor links, and raw anchors such as <a id="arch001"></a> or <a name="arch001"></a> become Confluence anchor macros instead of visible plain text.
Markdown or HTML tables are rewritten with Confluence full-width table layout attributes so wide stakeholder-facing tables use the available page width after publishing.
Before a real publish, the script checks Git for open changes to the source document, embedded local image or SVG diagram files, same-basename .drawio files for embedded SVG diagrams, and local Markdown links to .drawio or .svg diagrams. If any of those files have open changes, the script asks whether to commit those publish-related files before publishing. Choosing yes prompts for a commit message and commits only those publish-related files. Choosing no skips the commit and continues. After a successful publish, the script checks those files again and asks the same commit question only when Git reports remaining changes for those files. Use --skip-git-check only for automation where this prompt is intentionally bypassed.
If no Confluence page ID is found, the script always prompts the user to choose before credential loading, dry-run output, source updates, page creation, or publishing. The user can provide an existing Confluence link, or choose to create the page as a child of the overview page. The overview page is found by title in the target space and defaults to Overview; override this with --overview-title <title>.
When the script creates a new child page, it writes the returned Confluence link back to the source markdown, regenerates the content, and updates the new page so the Confluence copy contains the link too.
Markdown conversion uses the optional markdown package when installed. Without it, the script falls back to a small built-in converter for headings, paragraphs, pipe tables, lists, links, inline code, fenced code blocks, bold text, and italic text. For highly formatted documents, convert to storage-safe HTML before publishing.
Fenced markdown code blocks are converted to Confluence code macros. A language hint such as ```python is carried into the macro language parameter when present.
Use --content-format html to publish HTML directly, or --content-format markdown to force markdown conversion.