feat: add OpenCode rich document reader

.github/workflows/test.yml

@@ -0,0 +1,26 @@
+name: Test
+
+on:
+  push:
+  pull_request:
+
+permissions:
+  contents: read
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Check out repository
+        uses: actions/checkout@v4
+      - name: Set up Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: 22
+          cache: npm
+      - name: Install dependencies
+        run: npm ci
+      - name: Typecheck
+        run: npm run check
+      - name: Run tests
+        run: npm test

.gitignore

@@ -0,0 +1,4 @@
+node_modules/
+coverage/
+.roam/
+.DS_Store

AGENTS.md

@@ -0,0 +1,17 @@
+# Project agent memory
+
+This file is the project's committed home for project-intrinsic agent knowledge: build, test, release, architecture, and sharp-edge notes that should travel with the code.
+
+- Add durable project-specific notes here as they are discovered through real work.
+
+## Project notes
+
+- The published package entry point and OpenCode tool registration live in `src/index.ts`; the reader, safety checks, format registry, and media extraction are split across `src/`.
+- Run `npm run check && npm test` for the local typecheck and executable fixture suite. The DOCX, ODT, and PPTX fixtures are generated as real ZIP archives by `test/fixtures.ts`.
+
+## Maintaining this file
+
+Keep this file for knowledge useful to almost every future agent session in this project.
+Do not repeat what the codebase already shows; point to the authoritative file or command instead.
+Prefer rewriting or pruning existing entries over appending new ones.
+When updating this file, preserve this bar for all agents and keep entries concise.

CLAUDE.md

@@ -0,0 +1 @@
+AGENTS.md

README.md

@@ -22,11 +22,11 @@ Ask the agent to call `read_rich_document` with a document path inside the curre
 {"path":"docs/architecture.docx"}
 ```
 
-The tool returns structure-preserving Markdown for headings, lists, tables, links, sections, slide context, notes, and supported document content.
+The tool returns structure-preserving Markdown for headings, lists, tables, links, sections, slide context, notes, and supported document content. It also returns a structured media index in tool metadata.
 
 It appends an ordered embedded-media index containing each attachment's source name, MIME type, document section, and isolated temporary path.
 
-Use the built-in `read` tool on a listed image path when vision inspection is relevant.
+Use a media label from the index when vision inspection is relevant. The selected image is returned as a native OpenCode file attachment, while its isolated temporary path remains available for follow-up tools.
 
 Media is not attached by default, which prevents documents with many screenshots from inflating the model context.
 
@@ -36,6 +36,8 @@ To attach selected image media directly to the next model turn, call the tool ag
 {"path":"docs/architecture.docx","media":["media-1"]}
 ```
 
+Selectable image media supports `image/jpeg`, `image/png`, `image/gif`, `image/bmp`, `image/tiff`, `image/svg+xml`, and `image/webp` attachments.
+
 ## Safety and boundaries
 
 The source document is read without modification.