ai-wiki-xmind-ingest - SKILL.md Agent Skill

name: ai-wiki-xmind-ingest description: Export, digest, and ingest raw .xmind files for this AI Obsidian wiki. Use when processing XMind sources under the AI raw source root into wiki sources, concepts, entities, synthesis, maps, index, or log pages; when the user asks to run an XMind ingest; or when avoiding unnecessary inspect, tree, or validate steps during source conversion.

AI Wiki XMind Ingest

Use this skill whenever ingesting a raw .xmind file for this AI wiki. This skill owns XMind-specific ingest behavior; AGENTS.md owns global wiki rules, raw boundaries, source mapping, metadata, index/log, and question-page policy.

Required Context

Before writing, read AGENTS.md, docs/wiki-templates.md, and index.md. Do not bypass this skill with ad hoc xmind-cli commands; use the commands here as the approved path.

Fast Path

Pick or receive a raw .xmind under the raw source root from AGENTS.md.
Run the helper once. It must discover and export all sheets, not only the workbook default sheet:

python3 ".codex/skills/ai-wiki-xmind-ingest/scripts/export_xmind_source.py" "$RAW_XMIND" --json

Check ok and sheets_error. If ok is false, follow the Failure Path below instead of treating markdown as complete content.
Use source_relpath, target_source_note, raw metadata fields, sheet_count, sheets, exports, and combined markdown from the output to create or update the source note.
Treat per-sheet Markdown as temporary reading material. Read all exported sheets, understand them, then write an LLM-digested source note.
Update index.md, append one log.md line, then compile only reusable concept/entity/synthesis/map updates.

If the helper is unavailable, use this fallback sequence and still account for every sheet:

xmind sheets "$RAW_XMIND" --json
xmind export "$RAW_XMIND" --sheet-index 0 --format markdown
xmind export "$RAW_XMIND" --sheet-index 1 --format markdown
# ...repeat for every sheet index returned by xmind sheets

Do not rely on default xmind export "$RAW_XMIND" --format markdown for multi-sheet workbooks. A trailing xmind:#... link is only a sheet reference; it does not mean the linked sheet content was read.

XMind Source Rules

Use the source-note template from docs/wiki-templates.md. XMind-specific requirements:

source_type: xmind
status: ingested only when every sheet was discovered, exported, read, and digested.
status: partial when any sheet export is missing, failed, empty, or unreadable; record the affected sheet and next step.
status: blocked when sheet discovery/export cannot produce useful content.
Record sheet count and sheet titles in Source or Maintenance Notes; every sheet must be accounted for.
Source Digest must synthesize the workbook across sheets. Do not paste the complete exported Markdown, mechanical outline, first-N-lines sample, or truncated export.
Preserve reusable XMind details by digesting them into Source Digest, Key Claims, Links, or downstream pages: important branches, second-level taxonomies, critical paths, named frameworks, step lists, metric dimensions, placeholder/anomaly nodes, and numbering/structure problems. Do not write exported heading samples, mechanical outlines, or first-N-line snippets into Maintenance Notes as coverage proof.
Preserve every URL found in exported Markdown in External Links with brief context; if none are found, write No external links found in extracted content.
Use explicit: for claims directly stated or strongly supported by exported content; use inferred: for agent synthesis, classification, or implications.
Record placeholder titles, partial content, parse limits, and cleanup needs under Maintenance Notes.

Failure Path

Run diagnostics only when export fails, returns empty/garbled content, or the user asks for workbook metadata:

xmind inspect "$RAW_XMIND" --json
xmind validate "$RAW_XMIND" --json
xmind tree "$RAW_XMIND" --depth 4 --format markdown

If sheet discovery fails, do not treat a default export as complete. If diagnostics still fail, create a partial or blocked source note with attempted commands, error summary, and a concrete human next step. Continue with other sources when possible.

Do not create questions/ pages unless the user explicitly asks to save a question or answer.