Titan Frameworks (LangChain, LlamaIndex, Ollama, XRPL)
Provides token-efficient documentation search, latest syntax snippets, imports, migration notes, and error diagnosis for LangChain.
Provides token-efficient documentation search, latest syntax snippets, imports, migration notes, and error diagnosis for Ollama.
Provides token-efficient documentation search, latest syntax snippets, imports, migration notes, and error diagnosis for XRPL via xrpl.js and xrpl-py.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Titan Frameworks (LangChain, LlamaIndex, Ollama, XRPL) What's the correct xrpl.js syntax for issuing a token?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Titan Frameworks
Paid MCP that reviews and rewrites LangChain, LlamaIndex, Ollama, and XRPL code against current APIs — plus runnable examples, import lookup, and tx skeletons that never submit. Live docs are still included.
Live Streamable HTTP: https://titan-frameworks-production.up.railway.app/mcp
Paid with x402: $0.001 USDC on Base or 1000 drops XRP / 0.001 RLUSD on XRPL mainnet.
Agent discovery: llms.txt · Smithery smithery.yaml · MCP Registry server.json
Tools
Tool | Purpose |
| Lint a snippet for dead APIs before it crashes. Returns hunks, does not execute |
| Apply conservative catalog rewrites; leftover lists unsafe call sites |
| Stack-trace → deprecated API → exact fix. Use only when you have an error |
| Package, install, import, and call shape for one API name |
| Complete runnable file for a goal (agent, rag, payment, …) |
| Working snippets and migration notes for one topic. Use when writing a fragment |
| Keyword search over local Markdown. Use for concepts/symbols; not stack traces |
| Typed XRPL tx skeleton. Never submits or moves funds |
| Enumerate framework ids, aliases, and catalog topics (free |
| Full deprecation catalog for one framework or all four |
Prompts: migrate_framework_code, diagnose_stack_trace, start_from_example.
Pick one: source about to run → review_framework_code; need a patched file → rewrite_framework_code; stack trace → diagnose_framework_error; know the symbol → resolve_symbol; need a full file → fetch_working_example; topic fragment → fetch_latest_syntax; concept → search_ai_framework_docs; XRPL tx → draft_xrpl_intent_tx; which frameworks → list_supported_frameworks; all deprecations → list_known_deprecations.
Related MCP server: Atlas Docs MCP Server
Connect a client
{
"mcpServers": {
"titan-frameworks": {
"url": "https://titan-frameworks-production.up.railway.app/mcp"
}
}
}x402-capable clients send PAYMENT-SIGNATURE / X-PAYMENT after a 402. Local stdio (titan-frameworks / npm run dev) does not charge.
To pay production from Cursor, enable titan-frameworks-paid in MCP settings (runs scripts/cursor-x402-proxy.ts, signs Base USDC from EVM_PRIVATE_KEY in .env, cap X402_PROXY_MAX_CALLS default 25). Check the proxy with npx tsx scripts/cursor-x402-proxy.ts --check.
HTTP endpoints
Path | Notes |
| liveness + index stats |
| agent discovery |
| tool list + live pricing |
| x402scan resource index |
| full payment requirements + bazaar metadata |
| Smithery static card |
| Streamable HTTP MCP. Handshake/ |
Run locally (stdio)
cd titan-frameworks
npm install
npm run devProject config is in .cursor/mcp.json. Seed docs ship in src/data/seed-fallback.json.
npm run index-docsProduction re-indexes automatically when the live index is older than 24 hours. /admin can still trigger a crawl immediately.
HTTP mode (Vercel / Railway)
npm run dev:http
# POST http://127.0.0.1:3333/mcpnpm run build
npm startRailway uses railway.toml (npm run start). Vercel rewrites to api/index.ts.
x402 micropayments
Production billing is on. Unpaid tools/call on POST https://titan-frameworks-production.up.railway.app/mcp returns 402, except list_supported_frameworks which is free. initialize, ping, and tools/list are free so directories can health-check the connector.
Rail | Price | Network | Facilitator |
USDC | $0.001 | Base ( | PayAI |
XRP | 1000 drops (0.001 XRP) | XRPL mainnet ( | T54 |
RLUSD | 0.001 | XRPL mainnet ( | T54 |
A raw wallet transfer is not an x402 proof. Clients retry /mcp with PAYMENT-SIGNATURE / X-PAYMENT. Local stdio (npm run dev) does not charge.
To reproduce the gate locally: X402_ENABLED=true (mock) plus X402_USDC_LIVE=true / X402_XRPL_LIVE=true for live verify/settle.
Password-gated admin
Production dashboard: https://titan-frameworks-production.up.railway.app/admin
It is not a public stats page. Login is a password checked against ADMIN_SECRET_KEY. Unauthenticated /api/admin/* calls return 401. The old public /admin/api/stats route is gone (404).
Layout
src/app/ Next.js admin UI + /api/admin routes
src/components/ Admin dashboard React components
src/tools/ MCP tool handlers
src/mcp/tools.ts tool registration (includes xrpl)
src/indexer/ xrpl.js crawl targets
src/services/ search, indexer, store, x402, earnings, discovery
src/data/ JSON index, syntax catalog, known errors, symbol map, examples, XRPL intents
src/types/ Zod schemas + TS types
scripts/ index-docs, x402-smoke
server.json MCP Registry remote listing
smithery.yaml Smithery remote listing
glama.json Glama maintainer claimAvailable Tools
10 toolsdiagnose_framework_errorDiagnose Framework ErrorARead-onlyIdempotentInspect
Match a stack trace or exception against known deprecated APIs for langchain, llamaindex, ollama, or xrpl and return the replacement plus a fix. Use only when you have an error_log; matched=false means no catalog hit — then search_ai_framework_docs with the failing symbol. Lint source before it crashes with review_framework_code. Not for happy-path syntax (fetch_latest_syntax). Paid tools/call: $0.001 USDC or 1000 drops XRP; read-only, does not execute code.
Inputs: error_log (required), framework (required). Returns: hints, matched, matches, framework
draft_xrpl_intent_txDraft XRPL Intent TransactionARead-onlyIdempotentInspect
Draft a typed xrpl.js / xrpl-py skeleton for payment, trustline, rlusd, or channel, plus required fields and common tec/tem failure codes. Use when building an XRPL tx; never submits, signs, or moves funds. Prefer fetch_working_example for a full script and review_framework_code to lint Amount/address mistakes. Paid tools/call: $0.001 USDC or 1000 drops XRP.
Inputs: intent (required), language. Returns: hints, notes, intent, matched, +7 more
fetch_latest_syntaxFetch Latest SyntaxARead-onlyIdempotentInspect
Curated working imports, snippets, and migration notes for one topic (agents, rag, wallet, payment, trustlines, …). Use when writing or migrating a fragment; unknown topics fall back to related doc chunks instead of failing. Prefer fetch_working_example for a complete runnable file, search_ai_framework_docs for open-ended lookup, and diagnose_framework_error for exceptions. Paid tools/call: $0.001 USDC or 1000 drops XRP; read-only catalog.
Inputs: topic (required), framework (required). Returns: topic, imports, snippets, framework, +3 more
fetch_working_exampleFetch Working ExampleARead-onlyIdempotentInspect
Fetch one complete runnable file (filename, install/run, source) for a goal such as agent, rag, chat, payment, or rlusd. Use when fetch_latest_syntax snippets are too small to execute. matched=false means no catalog example — then fetch_latest_syntax. Paid tools/call: $0.001 USDC or 1000 drops XRP; does not execute the example.
Inputs: goal (required), framework (required), runtime, language. Returns: id, run, code, goals, +7 more
list_known_deprecationsList Known DeprecationsARead-onlyIdempotentInspect
Return the known-deprecated API catalog (id, old API, replacement, fix) for one framework or all four at once. Use when you want the full deprecation list without a stack trace or source snippet. Prefer diagnose_framework_error when you have an error_log and review_framework_code when you have code. Paid tools/call: $0.001 USDC or 1000 drops XRP; catalog-backed, does not execute code.
Inputs: framework. Returns: count, hints, framework, deprecations
list_supported_frameworksList Supported FrameworksARead-onlyIdempotentInspect
List the framework ids this server covers (langchain, llamaindex, ollama, xrpl) with display names, aliases, homepages, and catalog topics. Use when you do not know which framework string to pass. Free tools/call (no x402). Not a docs search (search_ai_framework_docs) and not a deprecation dump (list_known_deprecations). Catalog-backed.
Inputs: no named parameters. Returns: ids, count, frameworks
resolve_symbolResolve SymbolARead-onlyIdempotentInspect
Map an API name to the current package, install line, import, and call shape (ChatOpenAI, create_agent, VectorStoreIndex, xrpToDrops, …). Use when you know the symbol but not where it lives. Prefer search_ai_framework_docs for concepts and fetch_latest_syntax for topic snippets. Paid tools/call: $0.001 USDC or 1000 drops XRP; catalog-backed.
Inputs: symbol (required), framework (required). Returns: call, note, hints, symbol, +6 more
review_framework_codeReview Framework CodeARead-onlyIdempotentInspect
Scan source against known dead APIs (LLMChain, initialize_agent, ServiceContext, ripple-lib, …) and return each hit with replacement plus a unified-diff hunk. Use before running agent-written code; diagnose_framework_error is for after a stack trace. Does not execute or submit. Paid tools/call: $0.001 USDC or 1000 drops XRP; catalog-backed, not LLM-invented.
Inputs: code (required), framework (required), filename. Returns: hints, issues, matched, framework, +1 more
rewrite_framework_codeRewrite Framework CodeARead-onlyIdempotentInspect
Apply conservative catalog rewrites (imports/identifiers) to a snippet and return the patched file plus a full unified diff. Use when review_framework_code found hits and you want an applyable file. Leftover lists unsafe call-site transforms (e.g. LLMChain(...)) that stay for a human/agent. Does not execute or submit. Paid tools/call: $0.001 USDC or 1000 drops XRP.
Inputs: code (required), framework (required), filename. Returns: code, diff, notes, applied, +4 more
search_ai_framework_docsSearch AI Framework DocsARead-onlyIdempotentInspect
Keyword search over a local Markdown index for langchain, llamaindex, ollama, or xrpl; returns compact chunks and never live-crawls. Use for concepts, symbols, or questions; hitCount 0 means no match — shorten the query. Do not use for stack traces (diagnose_framework_error), import/package lookup (resolve_symbol), or copy-paste current syntax (fetch_latest_syntax). Paid tools/call: $0.001 USDC or 1000 drops XRP.
Inputs: query (required), framework (required). Returns: hits, query, hitCount, framework
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 10 tools
Each tool targets a distinct stage or granularity: source review, rewrite, error diagnosis, symbol resolution, examples, syntax, docs, XRPL drafting, framework listing, and deprecation listing. Descriptions explicitly steer between potentially overlapping tools, leaving little ambiguity.
All tool names follow a consistent snake_case verb_object pattern, with clear objects like framework_code, framework_error, symbol, working_example, and known_deprecations. Repeated verbs are paired with distinct objects, making the naming predictable.
Ten tools is well-scoped for a multi-framework migration and deprecation assistant. Each tool has a clear role, and no tool feels redundant or missing.
The toolset covers the full consult/refactor workflow: scan source, apply safe rewrites, diagnose errors, resolve symbols, fetch examples/syntax/docs, draft XRPL transactions, and enumerate framework/deprecation catalogs. Tools cross-reference next steps, so agents are not left at dead ends.
Maintenance
Related MCP Connectors
Access and search Branch's official API documentation directly within any AI agent or client.
Provide your AI coding tools with token-efficient access to up-to-date technical documentation for…
Token-efficient search for coding agents over public and private documentation.
@latest documentation and code examples to 9000+ libraries for LLMs and AI code editors in a singl…
Related MCP Servers
- FlicenseBqualityNot gradedmaintenanceSearches and fetches real-time documentation for libraries like Langchain, OpenAI, and Llama-Index using the Serper API. It allows LLMs to access up-to-date technical information and bypass knowledge cut-off limitations.1-
- AlicenseAqualityDmaintenanceProvides AI assistants with up-to-date documentation for popular libraries and frameworks, enabling them to generate more accurate code using less common or newly released libraries.536MIT
- FlicenseAqualityCmaintenanceEnables AI clients to search live documentation for popular developer libraries and return clean, source-cited summaries via a RAG pipeline.1-
- FlicenseNot gradedqualityAmaintenanceEnables AI models to search and retrieve up-to-date technical documentation for supported agent frameworks such as LangChain, LlamaIndex, AutoGen, Agno, OpenAI Agents SDK, MCP, Camel-AI, and CrewAI, returning extracted web page text over stdio or SSE.-