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 air-gapped/skills --skill vllm-reasoning-parsersgit clone --depth 1 https://github.com/air-gapped/skillsWrote 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/air-gapped/skills/vllm-reasoning-parsers)<a href="https://agentmods.dev/skills/air-gapped/skills/vllm-reasoning-parsers"><img src="https://agentmods.dev/badge/skills/air-gapped/skills/vllm-reasoning-parsers/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/air-gapped/skills/vllm-reasoning-parsers"><img src="https://agentmods.dev/badge/skills/air-gapped/skills/vllm-reasoning-parsers.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 | $0.00113 | $0.04444 |
| Opus 5 | $0.00056 | $0.02222 |
| Sonnet 5 | $0.00023 | $0.00889 |
| Haiku 4.5 | $0.00011 | $0.00444 |
Grade A, and why
vllm-reasoning-parsers 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 8d 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 — 169 lines — stays where its author put it; the contents beside it link to each section on GitHub.
vLLM reasoning parsers
Target: operators wiring up --reasoning-parser NAME on a chat-completion endpoint, or developers authoring a parser for a new thinking model. Source of truth: vllm/reasoning/ on main.
What a reasoning parser actually does
When a reasoning-trained model emits a single token stream like
<think>user asked X, let me check Y...</think>The answer is 42.
vLLM splits this into two fields on the chat-completion response: reasoning (the CoT) and content (the final answer). --reasoning-parser NAME selects the class that does the split. Without it, the whole stream lands in content.
Field-name note. On current
mainthe field isreasoning(seeChatMessage.reasoning/DeltaMessage.reasoninginvllm/entrypoints/openai/chat_completion/protocol.py). Pre-v0.19 code and many third-party docs / clients call itreasoning_content. If a client is readingreasoning_contentagainst a current-main server it will seenullevery time even when the parser ran correctly.
The parser is also the gating authority for xgrammar / structured output: by default, grammar enforcement is held off until is_reasoning_end(input_ids) flips true, so the model thinks freely before being constrained to JSON. Flip that default with --structured-outputs-config.enable_in_reasoning=true — then the grammar applies from token 0 regardless of reasoning state (useful for structured CoT).
The contract (ReasoningParser ABC)
vllm/reasoning/abs_reasoning_parsers.py. Every parser implements:
| Method | Called by | Purpose |
|---|---|---|
is_reasoning_end(input_ids) |
xgrammar, non-streaming serving, tool-call gate | Has </think>-equivalent been emitted yet? |
is_reasoning_end_streaming(input_ids, delta_ids) |
xgrammar per decode step | Same, but cheap — checks only the delta |
extract_content_ids(input_ids) -> list[int] |
structured output | Token IDs of post-reasoning content |
extract_reasoning(model_output, request) -> (reasoning, content) |
non-streaming chat completion | Full-string split; either field may be None |
extract_reasoning_streaming(previous_text, current_text, delta_text, previous_token_ids, current_token_ids, delta_token_ids) -> DeltaMessage | None |
streaming chat completion | Per-delta split; returns DeltaMessage(reasoning=..., content=...) or None to swallow |
What ships with it
7 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 8d ago First seen · 169 lines · 113 tokens per session scan A c6a2d4eb5c75
vllm-reasoning-parsers is a skill published in the GitHub repository air-gapped/skills (5 stars, last pushed 10d ago), licensed MIT. It adds 113 tokens to every session and 4,444 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 2026-09-03.
Other skills, from other repositories
serving-llms-vllm
Use when deploying production LLM APIs, optimizing inference latency/throughput, or serving models with limited GPU memory. Supports OpenAI-compatible endpoints, quantization (GPTQ/AWQ/FP8), and tensor parallelism.
omni-inference
The core OpenAI-compatible inference endpoints: chat completions, embeddings, images, audio (TTS/STT), moderations, rerank, and the Responses API. The primary integration surface for AI agents.
gemini-api-agent-platform
Guides the usage of the Gemini API on Agent Platform with the Google Gen AI SDK for enterprise AI applications. Covers SDK usage (Python, JS/TS, Go, Java, C#), capabilities like Live API, tools, multimedia generation, caching, and batch prediction.
sglang
Fast structured generation and serving for LLMs with RadixAttention prefix caching. Use for JSON/regex outputs, constrained decoding, agentic workflows with tool calls, or when you need 5× faster inference than vLLM with prefix sharing. Powers 300,000+ GPUs at xAI, AMD, NVIDIA, and LinkedIn.
llm-provider
Adds a new LLM provider implementing LLMProvider interface with call() and stream() methods. Integrates with provider factory in src/llm/index.ts, config detection in src/llm/config.ts, and error handling via tracking and recovery. Use when adding a new model backend, integrating a third-party LLM API, or extending…
calling-llms
Use when sending chat completions through liter-llm and routing to a specific provider via the provider/model prefix. Covers the chat call shape, provider routing, modelhint, message roles, and error categories.