Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add OrcaQubits/agentic-commerce-skills-plugins --skill nlweb-llm-providersgit clone --depth 1 https://github.com/OrcaQubits/agentic-commerce-skills-pluginsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/orcaqubits/agentic-commerce-skills-plugins/nlweb-llm-providers)<a href="https://agentmods.dev/skills/orcaqubits/agentic-commerce-skills-plugins/nlweb-llm-providers"><img src="https://agentmods.dev/badge/skills/orcaqubits/agentic-commerce-skills-plugins/nlweb-llm-providers/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/orcaqubits/agentic-commerce-skills-plugins/nlweb-llm-providers"><img src="https://agentmods.dev/badge/skills/orcaqubits/agentic-commerce-skills-plugins/nlweb-llm-providers.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00110 | $0.01923 |
| Opus 5 | $0.00055 | $0.00962 |
| Sonnet 5 | $0.00022 | $0.00385 |
| Haiku 4.5 | $0.00011 | $0.00192 |
Grade A, and why
nlweb-llm-providers scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 170 lines — stays where its author put it; the contents beside it link to each section on GitHub.
NLWeb LLM & Embedding Providers
Before writing code
Fetch live docs:
- Fetch https://github.com/nlweb-ai/NLWeb/blob/main/docs/nlweb-providers.md for the canonical provider list and config schema.
- Fetch https://github.com/nlweb-ai/NLWeb/blob/main/config/config_llm.yaml for the exact model IDs and env-var names currently shipped.
- Fetch https://github.com/nlweb-ai/NLWeb/blob/main/config/config_embedding.yaml for embedding defaults.
- Inspect
AskAgent/python/llm_providers/<provider>.pyfor the SDK calls the provider class makes. - Web-search the latest release notes — new providers and models get added often.
Conceptual Architecture
Mixed-Mode = Many Small LLM Calls
NLWeb's pipeline doesn't make one big LLM call per query. It makes many small calls: decontextualize the query, detect Schema.org item type, route to a tool, rank results, optionally summarize/generate. Each call has a strict <returnStruc> JSON schema in prompts.xml. Cost and latency are dominated by the number of calls, not the size of any single one.
High / Low Tier Model Selection
config_llm.yaml defines a high model and a low model per provider:
providers:
openai:
high: gpt-4.1
low: gpt-4.1-mini
api_key_env: OPENAI_API_KEY
The codebase decides which tier to use per call site — e.g., decontextualization is "low", final generate is "high". The exact assignment lives in core/ modules and the ModelRouter subsystem.
The Default Provider
Out of the box, NLWeb's preferred_endpoint (in config_llm.yaml) is azure_openai with gpt-4.1 / gpt-4.1-mini. Most users override this in .env or by editing the YAML.
All Supported LLM Providers
(Verify the live config_llm.yaml for current models and key names.)
| Provider | Default high | Default low | Env var |
|---|---|---|---|
| OpenAI | gpt-4.1 | gpt-4.1-mini | OPENAI_API_KEY |
| Azure OpenAI | gpt-4.1 | gpt-4.1-mini | AZURE_OPENAI_API_KEY + AZURE_OPENAI_ENDPOINT |
| Anthropic | claude-3-7-sonnet-latest | claude-3-5-haiku-latest | ANTHROPIC_API_KEY |
| Google Gemini | gemini-2.5-pro | gemini-2.0-flash-lite | GEMINI_API_KEY |
| DeepSeek on Azure | deepseek-coder-33b | deepseek-coder-7b | AZURE_DEEPSEEK_ENDPOINT |
| Llama on Azure | llama-2-70b | llama-2-13b | AZURE_LLAMA_ENDPOINT |
| HuggingFace | Qwen2.5-72B | Qwen2.5-Coder-7B | HF_TOKEN |
| Inception Labs | mercury-small | mercury-small | INCEPTION_API_KEY |
| Snowflake Cortex | claude-3-5-sonnet | llama3.1-8b | Snowflake creds |
| Ollama | configurable | configurable | local — no key |
| Pi Labs | (class present, may not be in default YAML) | — | — |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 170 lines · 110 tokens per session scan A e7a1e9585859
nlweb-llm-providers is a skill published in the GitHub repository OrcaQubits/agentic-commerce-skills-plugins (39 stars, last pushed 2d ago), licensed MIT. It adds 110 tokens to every session and 1,923 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-15.
Other skills, from other repositories
archestra-dev-llm-providers
Use when adding an LLM provider, changing proxy adapters or provider routes, fixing streaming/tool-call translation bugs, editing model fetchers or model handling, or touching provider credentials/enums and model constants.
chaingpt
Build with the ChainGPT Web3 AI developer platform. Full API/SDK reference and project scaffolding for: Web3 AI Chatbot & LLM, AI NFT Generator, Smart Contract Generator, Smart Contract Auditor, AI Crypto News, AgenticOS Twitter agents, and Solidity LLM. Use when building blockchain apps, Web3 chatbots, NFT tools…
switchboard
Complete Switchboard Oracle Protocol SDK for Solana - the permissionless oracle solution for price feeds, on-demand data, VRF randomness, and real-time streaming via Surge. Covers TypeScript SDK, Rust integration, Oracle Quotes, and all Switchboard tools.
0g-compute
0G Compute Network guide for decentralized AI inference, fine-tuning, and GPU services. Covers chatbots, image generation, speech-to-text, SDK integration (0g-serving-broker), processResponse API, broker.inference methods, CLI commands (0g-compute-cli), and account management. Use this skill for any 0G compute, 0G AI…
near-ai-cloud
NEAR AI Cloud private inference and verification. Use when integrating NEAR AI Cloud API for verifiable private AI inference, verifying model or gateway TEE attestation (NVIDIA NRAS, Intel TDX), verifying chat message signatures, implementing end-to-end encrypted chat, or using the OpenAI-compatible API with NEAR AI…
omni-compression
Configure RTK (command output), Caveman (prose), and stacked compression modes. Manage language packs, custom rules, and test prompt compression reducing tokens by 60–90%.