feat: items 7 + 9 bridge surfaces — 8 cartridges + boj_vector + boj_multimodal (#103)

## Summary

Closes the bridge-surface portion of epic #87 items 7 and 9. Backend
dispatch (Elixir/Zig) tracked separately in tracker issues #100 (vector)
and #101 (multi-modal).

Eight new cartridges in two domains; two new umbrella bridge tools.

## Item 7 — Vector databases (4 cartridges)

| Cartridge | Provider strength |
|---|---|
| \`pinecone-mcp\` | Hosted serverless indexes; dimension + metric
required at create-time |
| \`weaviate-mcp\` | Hybrid (vector + BM25) search; schema-driven
classes; modular vectorisers |
| \`qdrant-mcp\` | Rust-native; payload filtering; sparse + dense
vectors; self-host or cloud |
| \`chromadb-mcp\` | Embedded or client/server; LLM-app-focused;
document storage alongside vectors |

All four share a **7-operation manifest**: \`authenticate\` /
\`list_collections\` / \`create_collection\` / \`delete_collection\` /
\`upsert\` / \`query\` / \`delete\`. Provider-specific extras live under
\`params\` (Pinecone namespace, Weaviate alpha-hybrid weight, Qdrant
score_threshold, Chroma embedding_function).

## Item 9 — Multi-modal (4 cartridges)

| Cartridge | Role | Worker-compat? |
|---|---|---|
| \`whisper-mcp\` | Speech-to-text (OpenAI API + local whisper.cpp
backends) | ✅ HTTP API |
| \`elevenlabs-mcp\` | Text-to-speech, voice cloning (premium),
multilingual | ✅ HTTP API |
| \`replicate-mcp\` | Image/video/audio generation; async prediction
model | ✅ HTTP API |
| \`ffmpeg-mcp\` | Local transcoding (probe/extract/concat/trim) | ❌
**Local-only** per ADR-0013 |

ffmpeg-mcp is the glue: extract audio from video → whisper transcribe;
extract frames → replicate vision model; etc.

## Bridge tools (2 new)

Both follow the \`boj_search\` umbrella pattern from PR #99:

\`\`\`
boj_vector {operation, provider ∈ {pinecone|weaviate|qdrant|chromadb},
...}
boj_multimodal {operation, provider ∈
{whisper|elevenlabs|replicate|ffmpeg}, ...}
\`\`\`

Each dispatches via \`invokeCartridge\` to the right cartridge.
\`hardeningGate\` requires \`operation\`. Domain classifications added:
\`vector\`, \`multimodal\`.

## Offline-menu

All 8 cartridges added to \`mcp-bridge/lib/offline-menu.js\` (Teranga
tier) so they surface in \`boj_menu\` even when the REST backend is
offline.

## Tool count

42 (after \`boj_search\`) → **44** (after \`boj_vector\` +
\`boj_multimodal\`). Per the umbrella-tool design choice in #100/#101,
adding 8 cartridges added only 2 bridge tools — not 8.

## Tests

- ✅ 15/15 existing tests pass (no regressions)
- ✅ \`tools/list\` advertises both new tools end-to-end (verified via
\`deno run\`)

## What's NOT in this PR

- Idris2 ABI, Zig FFI, Deno adapter for any of the 8 cartridges — just
the manifest + bridge wiring
- Tests against live provider APIs — surface captured; contract testing
comes with backend impl
- Backend dispatch in the Elixir/Zig layer — per-cartridge follow-up
work (~1-2 days each per the trackers)

## Sequencing

This closes the **mechanical** portion of items 7 and 9. The trackers
(#100, #101) update on merge — backends remain open for follow-up
sessions.

Epic #87 status after this lands:
- Tier A: items 1, 8, 13 done in code; items 7, 9 **surfaces done**
(backends pending); item 14 RFC only
- Tier B: 6/6 RFCs done
- Tier C: items 11, 12 untouched

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: hyperpolymath

cartridges/chromadb-mcp/cartridge.json

@@ -0,0 +1,22 @@
+{
+  "$schema": "https://boj.dev/schemas/cartridge/v1.json",
+  "spdx": "MPL-2.0",
+  "copyright": "Copyright (c) 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk>",
+  "name": "chromadb-mcp",
+  "version": "0.1.0",
+  "description": "Chroma vector DB — embedded (local persistent) or client/server; LLM-app-focused; metadata + document storage alongside vectors.",
+  "domain": "vector",
+  "tier": "Teranga",
+  "protocols": ["MCP", "REST"],
+  "auth": { "method": "optional_bearer", "env_var": "CHROMA_AUTH_TOKEN", "credential_source": "Optional — embedded mode has no auth; client/server may use bearer token." },
+  "api": { "base_url": "local://chromadb-mcp", "content_type": "application/json" },
+  "tools": [
+    { "name": "vector_authenticate", "description": "Store Chroma endpoint URL (or 'embedded' for in-process) + optional bearer.", "inputSchema": { "type": "object", "properties": { "endpoint": { "type": "string" }, "auth_token": { "type": "string" }, "embedding_function": { "type": "string" } }, "required": ["endpoint"] } },
+    { "name": "vector_list_collections", "description": "List collections.", "inputSchema": { "type": "object", "properties": {} } },
+    { "name": "vector_create_collection", "description": "Create a collection with optional embedding function + metadata.", "inputSchema": { "type": "object", "properties": { "name": { "type": "string" }, "metadata": { "type": "object" }, "embedding_function": { "type": "string" } }, "required": ["name"] } },
+    { "name": "vector_delete_collection", "description": "Delete a collection.", "inputSchema": { "type": "object", "properties": { "name": { "type": "string" } }, "required": ["name"] } },
+    { "name": "vector_upsert", "description": "Add documents + vectors + metadata. Chroma computes embeddings if not provided.", "inputSchema": { "type": "object", "properties": { "collection": { "type": "string" }, "ids": { "type": "array" }, "documents": { "type": "array" }, "embeddings": { "type": "array" }, "metadatas": { "type": "array" } }, "required": ["collection", "ids"] } },
+    { "name": "vector_query", "description": "Query by text or embedding with where-filter + document-content filter.", "inputSchema": { "type": "object", "properties": { "collection": { "type": "string" }, "query_texts": { "type": "array" }, "query_embeddings": { "type": "array" }, "n_results": { "type": "integer" }, "where": { "type": "object" }, "where_document": { "type": "object" } }, "required": ["collection"] } },
+    { "name": "vector_delete", "description": "Delete documents by id or where-filter.", "inputSchema": { "type": "object", "properties": { "collection": { "type": "string" }, "ids": { "type": "array" }, "where": { "type": "object" } }, "required": ["collection"] } }
+  ]
+}

cartridges/elevenlabs-mcp/cartridge.json

@@ -0,0 +1,19 @@
+{
+  "$schema": "https://boj.dev/schemas/cartridge/v1.json",
+  "spdx": "MPL-2.0",
+  "copyright": "Copyright (c) 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk>",
+  "name": "elevenlabs-mcp",
+  "version": "0.1.0",
+  "description": "Text-to-speech via ElevenLabs API — high-quality voices, multilingual, voice cloning (premium tier), streaming output.",
+  "domain": "multimodal",
+  "tier": "Teranga",
+  "protocols": ["MCP", "REST"],
+  "auth": { "method": "api_key", "env_var": "ELEVENLABS_API_KEY", "credential_source": "ElevenLabs account → Profile → API Keys." },
+  "api": { "base_url": "local://elevenlabs-mcp", "content_type": "application/json" },
+  "tools": [
+    { "name": "audio_authenticate", "description": "Store ElevenLabs API key.", "inputSchema": { "type": "object", "properties": { "api_key": { "type": "string" } }, "required": ["api_key"] } },
+    { "name": "audio_synthesize", "description": "Generate speech from text. Returns audio (mp3 / pcm / opus).", "inputSchema": { "type": "object", "properties": { "text": { "type": "string" }, "voice_id": { "type": "string" }, "model_id": { "type": "string", "enum": ["eleven_multilingual_v2", "eleven_turbo_v2", "eleven_monolingual_v1"] }, "output_format": { "type": "string", "enum": ["mp3_44100_128", "pcm_44100", "opus_48000_128"] }, "stability": { "type": "number" }, "similarity_boost": { "type": "number" } }, "required": ["text", "voice_id"] } },
+    { "name": "audio_list_voices", "description": "List available voices on the account (preset + custom).", "inputSchema": { "type": "object", "properties": {} } },
+    { "name": "audio_clone_voice", "description": "Create a custom voice from sample audio (premium tier only).", "inputSchema": { "type": "object", "properties": { "name": { "type": "string" }, "samples": { "type": "array" }, "description": { "type": "string" } }, "required": ["name", "samples"] } }
+  ]
+}

cartridges/ffmpeg-mcp/cartridge.json

@@ -0,0 +1,21 @@
+{
+  "$schema": "https://boj.dev/schemas/cartridge/v1.json",
+  "spdx": "MPL-2.0",
+  "copyright": "Copyright (c) 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk>",
+  "name": "ffmpeg-mcp",
+  "version": "0.1.0",
+  "description": "Local FFmpeg gateway — probe metadata, transcode formats, extract audio, extract frames, concatenate, trim. Glue between whisper / replicate / browser screenshots. Local-only — requires host ffmpeg binary; not Worker-compatible.",
+  "domain": "multimodal",
+  "tier": "Teranga",
+  "protocols": ["MCP", "REST"],
+  "auth": { "method": "none", "env_var": null, "credential_source": "No auth — operates on local files. Sandbox via FFMPEG_ALLOWED_PATHS env if exposing to multi-tenant deployments." },
+  "api": { "base_url": "local://ffmpeg-mcp", "content_type": "application/json" },
+  "tools": [
+    { "name": "media_probe", "description": "Probe metadata (codec, duration, streams, dimensions). Read-only.", "inputSchema": { "type": "object", "properties": { "input": { "type": "string" } }, "required": ["input"] } },
+    { "name": "media_transcode", "description": "Convert between formats with optional bitrate / resolution / codec overrides.", "inputSchema": { "type": "object", "properties": { "input": { "type": "string" }, "output": { "type": "string" }, "video_codec": { "type": "string" }, "audio_codec": { "type": "string" }, "video_bitrate": { "type": "string" }, "resolution": { "type": "string" } }, "required": ["input", "output"] } },
+    { "name": "media_extract_audio", "description": "Extract audio track to its own file (mp3 / wav / opus). Useful as a Whisper input prep step.", "inputSchema": { "type": "object", "properties": { "input": { "type": "string" }, "output": { "type": "string" }, "format": { "type": "string", "enum": ["mp3", "wav", "opus", "aac"] } }, "required": ["input", "output"] } },
+    { "name": "media_extract_frames", "description": "Extract frames at a given interval or specific timestamps. Useful as Replicate vision-model input prep.", "inputSchema": { "type": "object", "properties": { "input": { "type": "string" }, "output_pattern": { "type": "string" }, "fps": { "type": "number" }, "timestamps": { "type": "array" } }, "required": ["input", "output_pattern"] } },
+    { "name": "media_concat", "description": "Concatenate multiple files into one (same codec required, or transcode-then-concat).", "inputSchema": { "type": "object", "properties": { "inputs": { "type": "array" }, "output": { "type": "string" } }, "required": ["inputs", "output"] } },
+    { "name": "media_trim", "description": "Trim/cut to a specific time range.", "inputSchema": { "type": "object", "properties": { "input": { "type": "string" }, "output": { "type": "string" }, "start": { "type": "string" }, "end": { "type": "string" }, "duration": { "type": "string" } }, "required": ["input", "output", "start"] } }
+  ]
+}

cartridges/pinecone-mcp/cartridge.json

@@ -0,0 +1,22 @@
+{
+  "$schema": "https://boj.dev/schemas/cartridge/v1.json",
+  "spdx": "MPL-2.0",
+  "copyright": "Copyright (c) 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk>",
+  "name": "pinecone-mcp",
+  "version": "0.1.0",
+  "description": "Pinecone hosted vector DB — serverless indexes, upsert, similarity search, namespaces, metadata filtering.",
+  "domain": "vector",
+  "tier": "Teranga",
+  "protocols": ["MCP", "REST"],
+  "auth": { "method": "api_key", "env_var": "PINECONE_API_KEY", "credential_source": "Pinecone console; environment-scoped." },
+  "api": { "base_url": "local://pinecone-mcp", "content_type": "application/json" },
+  "tools": [
+    { "name": "vector_authenticate", "description": "Store Pinecone API key.", "inputSchema": { "type": "object", "properties": { "api_key": { "type": "string" }, "environment": { "type": "string" } }, "required": ["api_key"] } },
+    { "name": "vector_list_collections", "description": "List indexes (Pinecone calls them indexes).", "inputSchema": { "type": "object", "properties": {} } },
+    { "name": "vector_create_collection", "description": "Create a new index. Pinecone-specific: dimension + metric required.", "inputSchema": { "type": "object", "properties": { "name": { "type": "string" }, "dimension": { "type": "integer" }, "metric": { "type": "string", "enum": ["cosine", "euclidean", "dotproduct"] } }, "required": ["name", "dimension"] } },
+    { "name": "vector_delete_collection", "description": "Delete an index.", "inputSchema": { "type": "object", "properties": { "name": { "type": "string" } }, "required": ["name"] } },
+    { "name": "vector_upsert", "description": "Insert/update vectors with optional metadata + namespace.", "inputSchema": { "type": "object", "properties": { "collection": { "type": "string" }, "vectors": { "type": "array" }, "namespace": { "type": "string" } }, "required": ["collection", "vectors"] } },
+    { "name": "vector_query", "description": "Similarity search. Supports metadata filter + namespace.", "inputSchema": { "type": "object", "properties": { "collection": { "type": "string" }, "query": { "type": "array" }, "top_k": { "type": "integer" }, "namespace": { "type": "string" }, "filter": { "type": "object" } }, "required": ["collection", "query"] } },
+    { "name": "vector_delete", "description": "Delete vectors by id, namespace, or filter.", "inputSchema": { "type": "object", "properties": { "collection": { "type": "string" }, "ids": { "type": "array" }, "namespace": { "type": "string" }, "filter": { "type": "object" } }, "required": ["collection"] } }
+  ]
+}

cartridges/qdrant-mcp/cartridge.json

@@ -0,0 +1,22 @@
+{
+  "$schema": "https://boj.dev/schemas/cartridge/v1.json",
+  "spdx": "MPL-2.0",
+  "copyright": "Copyright (c) 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk>",
+  "name": "qdrant-mcp",
+  "version": "0.1.0",
+  "description": "Qdrant vector DB — Rust-native; payloads + filtering; sparse + dense vectors; self-host or Qdrant Cloud.",
+  "domain": "vector",
+  "tier": "Teranga",
+  "protocols": ["MCP", "REST", "gRPC"],
+  "auth": { "method": "api_key", "env_var": "QDRANT_API_KEY", "credential_source": "Qdrant Cloud dashboard or self-hosted bearer." },
+  "api": { "base_url": "local://qdrant-mcp", "content_type": "application/json" },
+  "tools": [
+    { "name": "vector_authenticate", "description": "Store Qdrant URL + API key.", "inputSchema": { "type": "object", "properties": { "url": { "type": "string" }, "api_key": { "type": "string" } }, "required": ["url"] } },
+    { "name": "vector_list_collections", "description": "List collections.", "inputSchema": { "type": "object", "properties": {} } },
+    { "name": "vector_create_collection", "description": "Create a collection. Vector params: size (dim) + distance metric.", "inputSchema": { "type": "object", "properties": { "name": { "type": "string" }, "vector_size": { "type": "integer" }, "distance": { "type": "string", "enum": ["Cosine", "Euclid", "Dot"] } }, "required": ["name", "vector_size"] } },
+    { "name": "vector_delete_collection", "description": "Delete a collection.", "inputSchema": { "type": "object", "properties": { "name": { "type": "string" } }, "required": ["name"] } },
+    { "name": "vector_upsert", "description": "Insert/update points with payload.", "inputSchema": { "type": "object", "properties": { "collection": { "type": "string" }, "points": { "type": "array" } }, "required": ["collection", "points"] } },
+    { "name": "vector_query", "description": "Search by vector with payload filter + score threshold.", "inputSchema": { "type": "object", "properties": { "collection": { "type": "string" }, "vector": { "type": "array" }, "limit": { "type": "integer" }, "filter": { "type": "object" }, "score_threshold": { "type": "number" } }, "required": ["collection", "vector"] } },
+    { "name": "vector_delete", "description": "Delete points by id or filter.", "inputSchema": { "type": "object", "properties": { "collection": { "type": "string" }, "ids": { "type": "array" }, "filter": { "type": "object" } }, "required": ["collection"] } }
+  ]
+}

cartridges/replicate-mcp/cartridge.json

@@ -0,0 +1,20 @@
+{
+  "$schema": "https://boj.dev/schemas/cartridge/v1.json",
+  "spdx": "MPL-2.0",
+  "copyright": "Copyright (c) 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk>",
+  "name": "replicate-mcp",
+  "version": "0.1.0",
+  "description": "Replicate hosted ML models — image generation (Stable Diffusion, FLUX), video (Veo, Kling), upscaling, vision (LLaVA), audio (MusicGen). Async prediction model with polling.",
+  "domain": "multimodal",
+  "tier": "Teranga",
+  "protocols": ["MCP", "REST"],
+  "auth": { "method": "api_token", "env_var": "REPLICATE_API_TOKEN", "credential_source": "Replicate account → API tokens." },
+  "api": { "base_url": "local://replicate-mcp", "content_type": "application/json" },
+  "tools": [
+    { "name": "media_authenticate", "description": "Store Replicate API token.", "inputSchema": { "type": "object", "properties": { "api_token": { "type": "string" } }, "required": ["api_token"] } },
+    { "name": "media_run_model", "description": "Run a model by owner/name:version with inputs. Returns prediction_id; poll via media_get_prediction.", "inputSchema": { "type": "object", "properties": { "model": { "type": "string", "description": "Format owner/name or owner/name:version" }, "inputs": { "type": "object" }, "webhook": { "type": "string" } }, "required": ["model", "inputs"] } },
+    { "name": "media_list_models", "description": "Search or list models in the Replicate catalog by collection/owner/name.", "inputSchema": { "type": "object", "properties": { "query": { "type": "string" }, "owner": { "type": "string" }, "collection": { "type": "string" } } } },
+    { "name": "media_get_prediction", "description": "Get status + output of a prediction.", "inputSchema": { "type": "object", "properties": { "prediction_id": { "type": "string" } }, "required": ["prediction_id"] } },
+    { "name": "media_cancel_prediction", "description": "Cancel a running prediction.", "inputSchema": { "type": "object", "properties": { "prediction_id": { "type": "string" } }, "required": ["prediction_id"] } }
+  ]
+}

cartridges/weaviate-mcp/cartridge.json

@@ -0,0 +1,22 @@
+{
+  "$schema": "https://boj.dev/schemas/cartridge/v1.json",
+  "spdx": "MPL-2.0",
+  "copyright": "Copyright (c) 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk>",
+  "name": "weaviate-mcp",
+  "version": "0.1.0",
+  "description": "Weaviate vector DB — hybrid (vector + BM25 + filter) search, schema-driven classes, modular vectorisers; self-host or cloud.",
+  "domain": "vector",
+  "tier": "Teranga",
+  "protocols": ["MCP", "REST", "gRPC", "GraphQL"],
+  "auth": { "method": "api_key", "env_var": "WEAVIATE_API_KEY", "credential_source": "Weaviate Cloud Console or self-hosted instance bearer." },
+  "api": { "base_url": "local://weaviate-mcp", "content_type": "application/json" },
+  "tools": [
+    { "name": "vector_authenticate", "description": "Store Weaviate endpoint URL + API key.", "inputSchema": { "type": "object", "properties": { "endpoint": { "type": "string" }, "api_key": { "type": "string" } }, "required": ["endpoint"] } },
+    { "name": "vector_list_collections", "description": "List classes (Weaviate calls collections 'classes').", "inputSchema": { "type": "object", "properties": {} } },
+    { "name": "vector_create_collection", "description": "Create a class with property schema + vectoriser.", "inputSchema": { "type": "object", "properties": { "name": { "type": "string" }, "properties": { "type": "array" }, "vectorizer": { "type": "string" } }, "required": ["name"] } },
+    { "name": "vector_delete_collection", "description": "Delete a class and all its objects.", "inputSchema": { "type": "object", "properties": { "name": { "type": "string" } }, "required": ["name"] } },
+    { "name": "vector_upsert", "description": "Insert/update objects; vectoriser computes embedding if not provided.", "inputSchema": { "type": "object", "properties": { "collection": { "type": "string" }, "objects": { "type": "array" } }, "required": ["collection", "objects"] } },
+    { "name": "vector_query", "description": "Hybrid (vector + BM25 + filter) search. Set alpha to weight vector vs BM25.", "inputSchema": { "type": "object", "properties": { "collection": { "type": "string" }, "query": { "type": "string" }, "vector": { "type": "array" }, "alpha": { "type": "number" }, "limit": { "type": "integer" }, "where": { "type": "object" } }, "required": ["collection"] } },
+    { "name": "vector_delete", "description": "Delete objects by id or where-filter.", "inputSchema": { "type": "object", "properties": { "collection": { "type": "string" }, "ids": { "type": "array" }, "where": { "type": "object" } }, "required": ["collection"] } }
+  ]
+}

cartridges/whisper-mcp/cartridge.json

@@ -0,0 +1,19 @@
+{
+  "$schema": "https://boj.dev/schemas/cartridge/v1.json",
+  "spdx": "MPL-2.0",
+  "copyright": "Copyright (c) 2026 Jonathan D.A. Jewell (hyperpolymath) <j.d.a.jewell@open.ac.uk>",
+  "name": "whisper-mcp",
+  "version": "0.1.0",
+  "description": "Speech-to-text via OpenAI Whisper API + local whisper.cpp fallback. Transcription, language detection, optional translation to English.",
+  "domain": "multimodal",
+  "tier": "Teranga",
+  "protocols": ["MCP", "REST"],
+  "auth": { "method": "optional_api_key", "env_var": "OPENAI_API_KEY", "credential_source": "Required for API backend; local whisper.cpp backend needs no auth." },
+  "api": { "base_url": "local://whisper-mcp", "content_type": "application/json" },
+  "tools": [
+    { "name": "audio_authenticate", "description": "Store OpenAI key (for API backend) or path to whisper.cpp binary (for local).", "inputSchema": { "type": "object", "properties": { "backend": { "type": "string", "enum": ["openai", "local"] }, "api_key": { "type": "string" }, "model_path": { "type": "string" } }, "required": ["backend"] } },
+    { "name": "audio_transcribe", "description": "Transcribe audio file or URL to text. Word-level timestamps optional.", "inputSchema": { "type": "object", "properties": { "source": { "type": "string" }, "language": { "type": "string" }, "timestamps": { "type": "boolean" }, "model": { "type": "string", "enum": ["whisper-1", "tiny", "base", "small", "medium", "large", "large-v3"] } }, "required": ["source"] } },
+    { "name": "audio_detect_language", "description": "Detect spoken language from a short audio sample.", "inputSchema": { "type": "object", "properties": { "source": { "type": "string" } }, "required": ["source"] } },
+    { "name": "audio_translate", "description": "Transcribe and translate any source language to English.", "inputSchema": { "type": "object", "properties": { "source": { "type": "string" }, "model": { "type": "string" } }, "required": ["source"] } }
+  ]
+}