Release 0.3.1

Changed

• Rebrand: aipartnerup → aiperceivable

Release 0.3.0

Release version 0.3.0

Release 0.2.1

Release version 0.2.1

Release 0.2.0

Changed

• BREAKING: Upgraded apcore-js peer dependency from ^0.6.0 to ^0.13.0.
• BREAKING: Upgraded apcore-mcp peer dependency from ^0.7.0 to ^0.10.0.
• BREAKING: Removed ExplorerHandler and ExplorerHandlerOptions re-exports from tiptap-apcore/server — these no longer exist in apcore-mcp 0.10.0 (replaced by mcp-embedded-ui).
• Query command annotations now include cacheable: true to leverage apcore-mcp 0.10.0 caching support.

Added

• Re-export asyncServe() from tiptap-apcore/server — embeddable MCP HTTP handler for mounting into existing servers (new in apcore-mcp 0.9.0).
• Re-export AsyncServeOptions and AsyncServeApp types from both tiptap-apcore and tiptap-apcore/server.

Dependencies

• Peer: apcore-js ^0.13.0 (was ^0.6.0), apcore-mcp ^0.10.0 (was ^0.7.0).

Release 0.1.0

Initial release of tiptap-apcore — let AI safely control your TipTap editor through standardized, schema-validated, access-controlled modules.

Added

Core API

• withApcore(editor, options?) factory function — single entry point that returns an APCore Registry + Executor from any TipTap editor instance.
• Configurable options: acl, prefix, includeUnsafe, logger, sanitizeHtml.
• Server-only exports via tiptap-apcore/server: serve(), toOpenaiTools(), resolveRegistry(), resolveExecutor(), ExplorerHandler.

Command Coverage — 79 Built-in Commands across 6 Categories

• Query (10): getHTML, getJSON, getText, isActive, getAttributes, isEmpty, isEditable, isFocused, getCharacterCount, getWordCount — read-only, idempotent.
• Format (36): toggleBold, toggleItalic, toggleStrike, toggleCode, toggleUnderline, toggleSubscript, toggleSuperscript, toggleHighlight, toggleHeading, toggleBulletList, toggleOrderedList, toggleTaskList, toggleCodeBlock, toggleBlockquote, setTextAlign, setBold, setItalic, setStrike, setCode, setLink, setMark, setHeading, setParagraph, setBlockquote, setHardBreak, setHorizontalRule, unsetBold, unsetItalic, unsetStrike, unsetCode, unsetLink, unsetMark, unsetAllMarks, unsetBlockquote, clearNodes, updateAttributes.
• Content (15): insertContent, insertContentAt, setNode, splitBlock, liftListItem, sinkListItem, wrapIn, joinBackward, joinForward, lift, splitListItem, wrapInList, toggleList, exitCode, deleteNode.
• Destructive (6): clearContent, setContent, deleteSelection, deleteRange, deleteCurrentNode, cut — marked destructive: true, requiresApproval: true.
• Selection (10): setTextSelection, setNodeSelection, selectAll, selectParentNode, selectTextblockStart, selectTextblockEnd, selectText, focus, blur, scrollIntoView.
• History (2): undo, redo.

Custom APCore Command: selectText

• Semantic text search and selection — not a native TipTap command.
• Parameters: text (required string), occurrence (optional, 1-based integer, defaults to 1).
• Returns { found: boolean, from?: number, to?: number }.
• Substring matching via doc.descendants() traversal.

Selection Effect Metadata

• New SelectionEffect type (require | preserve | destroy | none) added to AnnotationEntry.
• Every command annotated with its selection behavior at the source level (AnnotationCatalog).
• Exposed via ModuleDescriptor.metadata.selectionEffect for external consumers to read dynamically.

Automatic Extension Discovery

• ExtensionScanner introspects editor.extensionManager.extensions at runtime.
• Discovers commands from addCommands() on each extension.
• Detects node, mark, and extension types.
• Dynamic re-discovery via registry.discover() — only emits events for changed modules.

Module Builder & Schema Catalog

• ModuleBuilder transforms command names into complete ModuleDescriptor objects.
• Module ID pattern: {prefix}.{category}.{commandName} (e.g., tiptap.format.toggleBold).
• SchemaCatalog provides strict JSON Schema definitions for all 79 commands — additionalProperties: false for OpenAI strict mode compatibility.
• Human-readable descriptions auto-generated from camelCase names.
• Documentation links auto-generated: https://tiptap.dev/docs/editor/api/commands/{commandName}.

Safety Annotations

• 6 safety flags per command: readonly, destructive, idempotent, requiresApproval, openWorld, streaming.
• AnnotationCatalog provides static metadata for all known commands.
• Unknown commands excluded by default (includeUnsafe: false).

Role-Based Access Control (ACL)

• AclGuard with 3 preset roles:
• readonly — query commands only.
• editor — query + format + content + history + selection.
• admin — all categories including destructive.
• Fine-grained overrides: allowTags, denyTags, allowModules, denyModules.
• Precedence: denyModules > allowModules > denyTags > allowTags > role defaults.
• Detailed denial reasons for diagnostics.

Security Hardening

• URL protocol validation for setLink — allows http, https, mailto, tel, and relative paths.
• Prototype pollution guard — blocks __proto__, constructor, prototype in property names.
• Optional sanitizeHtml callback for insertContent, insertContentAt, setContent.
• Input validation: non-empty strings, finite numbers, required field enforcement.

Error Handling

• TiptapModuleError with 7 error codes: MODULE_NOT_FOUND, COMMAND_NOT_FOUND, SCHEMA_VALIDATION_ERROR, ACL_DENIED, EDITOR_NOT_READY, COMMAND_FAILED, INTERNAL_ERROR.
• Errors include moduleId, code, and contextual details.

Registry & Event System

• TiptapRegistry with full module lifecycle: register(), unregister(), list(), getDefinition(), iter().
• Filtering by tags (OR logic) and module ID prefix.
• Event system: on('register'), on('unregister') with cleanup via off() and removeAllListeners().

Integration Exports

• MCP Server via serve(executor) — supports stdio, HTTP streaming, SSE transports.
• OpenAI Function Calling via toOpenaiTools(executor).
• ExplorerHandler for interactive module discovery UI.

Demo Application

• Multi-Provider LLM Support: OpenAI (GPT-4o, GPT-4.1, GPT-5.1), Anthropic (Claude Sonnet 4.5, Haiku 4.5, Opus 4.5), Google Gemini (2.5 Flash, 2.5 Pro, 2.0 Flash).
• Dynamic System Prompt: Built at request time from registry metadata — command list, selection behavior classification, task patterns, anti-patterns. No hardcoded command names.
• Gemini Schema Sanitization: Deep-clone sanitizer that removes additionalProperties, converts type arrays, flattens anyOf, filters required, and preserves empty object schemas from SDK stripping.
• Headless TipTap Editor: JSDOM-based server-side editor for AI command execution, created and destroyed per request.
• Frontend Undo/Version Stack: useRef-based HTML history buffer (max 50 entries) with undo and clear controls.
• Model Selector: Grouped dropdown populated from /api/health with per-provider API key detection.
• Chat UI: Multi-turn conversation, tool call inspection (expandable details), loading animation, keyboard shortcuts.
• Demo Scenarios: AI-driven insert, format, and clear with confirmation dialog for destructive operations.
• Tool Panel: Registry-driven command browser with category badges and manual execution.
• ACL Switcher: Live role switching with APCore re-initialization.
• Responsive Layout: 4-column grid with breakpoints at 1400px and 1024px.

Test Suite

• 925 tests across 10 test files (9 unit, 1 integration).
• Coverage: unit tests for every public class and function.
• Components tested: TiptapModuleError, AnnotationCatalog, SchemaCatalog, ModuleBuilder, ExtensionScanner, TiptapRegistry, AclGuard, TiptapExecutor, withApcore, integration workflows.

Dependencies

• Peer: @tiptap/core ^2.0.0, apcore-js ^0.6.0, apcore-mcp ^0.7.0.
• Runtime: Node.js >= 18.0.0.
• License: Apache-2.0.