Find Simplifications
vllm-project/vllm-omni
Find evidence-backed simplification candidates in vLLM-Omni and, when requested, turn them into focused proposals or code changes.
Review PRs and local changes for hsliuustc0106/vllm-rlt: Ouro engine and KV correctness, serving behavior, and BF16 accuracy/speed evidence.
Install Vllm Rlt Review PR
$ npx skills add ThinkFlowLab/vllm-rlt --skill vllm-rlt-review-pr -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install ThinkFlowLab/vllm-rlt vllm-rlt-review-pr --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/ThinkFlowLab/vllm-rlt.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/vllm-rlt-review-pr .claude/skills/vllm-rlt-review-pr && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "vllm-rlt-review-pr" agent skill from https://github.com/ThinkFlowLab/vllm-rlt/tree/main/.agents/skills/vllm-rlt-review-pr into .claude/skills/vllm-rlt-review-pr/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "vllm-rlt-review-pr", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/ThinkFlowLab/vllm-rlt/tree/main/.agents/skills/vllm-rlt-review-prType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add ThinkFlowLab/vllm-rlt --skill vllm-rlt-review-pr -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install ThinkFlowLab/vllm-rlt vllm-rlt-review-pr --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/ThinkFlowLab/vllm-rlt.git skills-src && mkdir -p .agents/skills && cp -r skills-src/.agents/skills/vllm-rlt-review-pr .agents/skills/vllm-rlt-review-pr && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "vllm-rlt-review-pr" agent skill from https://github.com/ThinkFlowLab/vllm-rlt/tree/main/.agents/skills/vllm-rlt-review-pr into .agents/skills/vllm-rlt-review-pr/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "vllm-rlt-review-pr", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add ThinkFlowLab/vllm-rlt --skill vllm-rlt-review-pr -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install ThinkFlowLab/vllm-rlt vllm-rlt-review-pr --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/ThinkFlowLab/vllm-rlt.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/.agents/skills/vllm-rlt-review-pr .cursor/skills/vllm-rlt-review-pr && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "vllm-rlt-review-pr" agent skill from https://github.com/ThinkFlowLab/vllm-rlt/tree/main/.agents/skills/vllm-rlt-review-pr into .cursor/skills/vllm-rlt-review-pr/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "vllm-rlt-review-pr", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/ThinkFlowLab/vllm-rlt.git --path .agents/skills/vllm-rlt-review-pr--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add ThinkFlowLab/vllm-rlt --skill vllm-rlt-review-pr -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install ThinkFlowLab/vllm-rlt vllm-rlt-review-pr --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/ThinkFlowLab/vllm-rlt.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/.agents/skills/vllm-rlt-review-pr .gemini/skills/vllm-rlt-review-pr && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "vllm-rlt-review-pr" agent skill from https://github.com/ThinkFlowLab/vllm-rlt/tree/main/.agents/skills/vllm-rlt-review-pr into .gemini/skills/vllm-rlt-review-pr/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "vllm-rlt-review-pr", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install ThinkFlowLab/vllm-rlt vllm-rlt-review-prInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add ThinkFlowLab/vllm-rlt --skill vllm-rlt-review-pr -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/ThinkFlowLab/vllm-rlt.git skills-src && mkdir -p .github/skills && cp -r skills-src/.agents/skills/vllm-rlt-review-pr .github/skills/vllm-rlt-review-pr && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "vllm-rlt-review-pr" agent skill from https://github.com/ThinkFlowLab/vllm-rlt/tree/main/.agents/skills/vllm-rlt-review-pr into .github/skills/vllm-rlt-review-pr/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "vllm-rlt-review-pr", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add ThinkFlowLab/vllm-rlt --skill vllm-rlt-review-pr -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install ThinkFlowLab/vllm-rlt vllm-rlt-review-pr --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/ThinkFlowLab/vllm-rlt.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/.agents/skills/vllm-rlt-review-pr .opencode/skills/vllm-rlt-review-pr && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "vllm-rlt-review-pr" agent skill from https://github.com/ThinkFlowLab/vllm-rlt/tree/main/.agents/skills/vllm-rlt-review-pr into .opencode/skills/vllm-rlt-review-pr/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "vllm-rlt-review-pr", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
vllm-rlt-review-pr- GitHub stars
- 149
- Token cost
- ~1.1k tokens
- SKILL.md length
- 540 words
- Files
- 6 (incl. references)
- Skills in repo
- 3
- Repo updated
- First seen
- Licence
- Apache-2.0
At a glance
Review PRs and local changes for hsliuustc0106/vllm-rlt: Ouro engine and KV correctness, serving behavior, and BF16 accuracy/speed evidence.
- Maintainer reviews
- SKILL.md covers Freeze the review, Load relevant references and Validate and deliver
- Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md
- Author self-reviews of this project
What it does
Vllm Rlt Review PR is an agent skill from ThinkFlowLab/vllm-rlt. Review PRs and local changes for hsliuustc0106/vllm-rlt: Ouro engine and KV correctness, serving behavior, and BF16 accuracy/speed evidence. Use for maintainer reviews, repeat reviews, and author self-reviews of this project; not for vLLM-Omni or unrelated repositories.
Its SKILL.md is about 1.1k tokens, which your agent loads only when the skill is triggered. The skill folder holds 7 other files, including reference files (for example `agents/openai.yaml`, `references/ab-evidence.md` and `references/ab-example.md`).
It sits in AI & LLM Engineering, covering LLM inference and serving and Pull requests. It works with vLLM. The repository describes itself as: vllm based inference engine for looped transformer. The licence is Apache-2.0.
When your agent uses it
- Maintainer reviews
- Author self-reviews of this project
- Not for vLLM-Omni
- Unrelated repositories
Example prompts
- “/vllm-rlt-review-pr”
What it can do on your machine
Read from SKILL.md and the folder at commit b599dc5. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
No scripts in the folder and no shell commands in SKILL.md.
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
Vllm Rlt Review PR loads about 1.1k tokens when it runs, and up to ~8.6k if it reads all its reference files. Until then it costs about 72 tokens; SKILL.md has 540 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
SKILL.md
The full file from ThinkFlowLab/vllm-rlt at commit b599dc5, republished under its Apache-2.0 licence (© ThinkFlowLab). 540 words, ~1,084 tokens.
Download SKILL.mdSave it as
.claude/skills/vllm-rlt-review-pr/SKILL.md (or your agent's skills folder). This skill also uses 5 other files; get the full folder from GitHub.- name
- vllm-rlt-review-pr
- description
- Review PRs and local changes for hsliuustc0106/vllm-rlt: Ouro engine and KV correctness, serving behavior, and BF16 accuracy/speed evidence. Use for maintainer reviews, repeat reviews, and author self-reviews of this project; not for vLLM-Omni or unrelated repositories.
Review vllm-rlt changes
Review the requested change against its intended base. Return actionable, evidence-backed defects; zero findings is valid. Distinguish implementation correctness, reference fidelity, accuracy, speed and memory qualification.
Freeze the review
- Verify the repository/remotes and applicable AGENTS.md. The project is
hsliuustc0106/vllm-rlt; accept its verified forks. Do not inherit vLLM-Omni's architecture, reviewers, test requirements or milestone gates. - For a PR, record base/head SHAs, merge base, state, description, full diff, relevant review threads and checks. Read code from the frozen head, not a moving local branch. A stacked PR's prerequisite stack is its immediate comparison base; distinguish a cumulative comparison against main.
- For local changes, record HEAD, intended base, status and the relevant staged, unstaged and untracked diff. Preserve user changes. A review does not authorize source fixes, commits, benchmark campaigns or posting a GitHub review.
- Before delivery, check whether the PR head changed. Label the reviewed SHA; recheck affected findings if continuing against the newer head. Do not present old CI or experiment evidence as validation of an untested newer revision.
- When a prerequisite is reported merged, fetch the verified upstream main before selecting a new experiment base. If PR metadata disagrees with Git, verify the merge's parents/ancestry or integrated diff rather than relying on a cached API state or commit subject alone. Report the discrepancy and the verified Git SHA; do not keep treating an integrated prerequisite as open. Preserve the requested snapshot for an ongoing review.
Load relevant references
Read the diff and its callers to select references. Load only those relevant to the changed behavior or claims, and apply their checks proportionally.
| Reference | Read when |
|---|---|
| Test quality and review execution | Production or test changes, executable validation, or repeat reviews. Covers regression evidence, source identity, simplification and revalidation. |
| Core and interface review | Core behavior or public API, CLI, configuration, outputs or UI changes. Covers why the change is necessary, Plan B, invariants and compatibility. |
| A/B evidence | Inference, numerics, memory, measurement logic or accuracy/performance claims are affected. Covers applicable BF16 checks and frozen experiment gates. |
No A/B is required for unaffected runtime/measurement paths. Correctness-only changes need relevant regression checks, not a speedup.
Show full SKILL.md (187 more words)Show less
Validate and deliver
Use targeted CPU tests or a small reproducer to resolve a concrete uncertainty. Check test dependencies and skips; a green suite with GPU cases skipped does not validate kernels or real serving. GPU validation requires the user's applicable experiment scope and verified scheduler reservation; never use an unreserved device or silently expand a benchmark's exhausted run budget. If unavailable, finish the code/evidence review and identify the remaining validation precisely.
For each finding, give priority, a short defect title, the smallest relevant changed-line location, a reachable trigger, observable impact and supporting evidence. Trace the failure into unchanged callers if needed, but attribute it to this change. Exclude speculative risks, unrelated backlog and style-only feedback. In repeat reviews, verify fixes at the new snapshot and avoid reposting resolved findings.
Follow the applicable reference's reporting emphasis, then give findings in priority order and a brief validation/scope note with the reviewed SHA. If there are no findings, say so without implying unrun gates passed. Use verified PR diff links or absolute local file links. Draft/post comments only within the user's explicit authorization; select reviewers from actual ownership evidence.
© ThinkFlowLab, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
SKILL.md and 5 other files (references) in .agents/skills/vllm-rlt-review-pr of ThinkFlowLab/vllm-rlt.
- SKILL.md
- agents/openai.yaml
- references/ab-evidence.md
- references/ab-example.md
- references/core-and-interface.md
- references/test-quality-and-execution.md
Open the folder on GitHubat commit b599dc5
Compare with similar skills
Vllm Rlt Review PR next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Vllm Rlt Review PR this skillThinkFlowLab/vllm-rlt | 149 | — | ~1.1k | Automated safety check: Pass | Apache-2.0 | |
| Find Simplificationsvllm-project/vllm-omni | 7.1k | — | ~2.7k | Automated safety check: Pass | Apache-2.0 | |
| Review PRvllm-project/vllm-omni | 7.1k | — | ~3.8k | Automated safety check: Pass | Apache-2.0 | |
| SageMaker Serving Image Selectionhuggingface/skills | 11k | 1 repos | ~4.6k | Automated safety check: Pass | Apache-2.0 | |
| Hugging Face Local Model Evalshuggingface/skills | 11k | 2 repos | ~1.6k | Automated safety check: Pass | Apache-2.0 | |
| CI Fails Buildkiteguqiong96/Lvllm | 465 | 2 repos | ~349 | Automated safety check: Pass | Apache-2.0 |
Similar skills
Review PR
vllm-project/vllm-omni
Review pull requests and local branches for vllm-project/vllm-omni with a frozen snapshot, module-design ownership, feature-design overlays, targeted validation, and concise evidence-backed findings.
- Official
SageMaker Serving Image Selection
huggingface/skills
Chooses the right serving container and current image URI for deploying a Hugging Face model to a SageMaker endpoint, preferring Hugging Face images over generic ones.
- Official
Hugging Face Local Model Evals
huggingface/skills
Runs evaluations of Hugging Face Hub models on local hardware with inspect-ai or lighteval, and helps choose between vLLM, Transformers and accelerate backends.
CI Fails Buildkite
guqiong96/Lvllm
Fetch and diagnose vLLM Buildkite CI failure logs. An agent skill from guqiong96/Lvllm.
Gptqmodel Tokenizer Normalization
ModelCloud/GPTQModel
Diagnose and correct GPT-QModel tokenizer initialization, tokenization normalization, special-token handling, prompt rendering, and chat-template problems.
More from ThinkFlowLab/vllm-rlt
Rlt Refactor
ThinkFlowLab/vllm-rlt
Review and refactor inference-runtime code using concrete rules for responsibility boundaries, state ownership, interfaces, asynchronous lifetimes, KV management, and maintainability.
Rlt Perf Opt
ThinkFlowLab/vllm-rlt
Analyze and optimize vllm-rlt inference performance using reproducible unprofiled benchmarks, paired ops-only/full profiles, source-level attribution, and correctness checks.
Related
Works with
Categories
Questions about Vllm Rlt Review PR
What does Vllm Rlt Review PR do?
Review PRs and local changes for hsliuustc0106/vllm-rlt: Ouro engine and KV correctness, serving behavior, and BF16 accuracy/speed evidence. Vllm Rlt Review PR is an agent skill from ThinkFlowLab/vllm-rlt. Review PRs and local changes for hsliuustc0106/vllm-rlt: Ouro engine and KV correctness, serving behavior, and BF16 accuracy/speed evidence.
When should I use Vllm Rlt Review PR?
Vllm Rlt Review PR fits situations like: maintainer reviews; author self-reviews of this project; not for vLLM-Omni; unrelated repositories.
How do I install Vllm Rlt Review PR in Claude Code?
Run `npx skills add ThinkFlowLab/vllm-rlt --skill vllm-rlt-review-pr -a claude-code`. Or copy the skill folder (.agents/skills/vllm-rlt-review-pr in ThinkFlowLab/vllm-rlt) into .claude/skills/vllm-rlt-review-pr in your project. Claude Code loads it when a task matches its description.
How do I install Vllm Rlt Review PR in Codex?
Run `npx skills add ThinkFlowLab/vllm-rlt --skill vllm-rlt-review-pr -a codex`. Or copy the skill folder (.agents/skills/vllm-rlt-review-pr in ThinkFlowLab/vllm-rlt) into .agents/skills/vllm-rlt-review-pr in your project. Codex loads it when a task matches its description.
Can I use Vllm Rlt Review PR in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add ThinkFlowLab/vllm-rlt --skill vllm-rlt-review-pr -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/vllm-rlt-review-pr, .gemini/skills/vllm-rlt-review-pr, .github/skills/vllm-rlt-review-pr and .opencode/skills/vllm-rlt-review-pr in your project.
What does Vllm Rlt Review PR need to run?
SKILL.md names no scripts, command-line tools or credentials: Vllm Rlt Review PR is instructions for the agent only.
Does Vllm Rlt Review PR access the network?
SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.
Is Vllm Rlt Review PR safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
What licence does Vllm Rlt Review PR use?
Vllm Rlt Review PR is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Vllm Rlt Review PR use?
About 1.1k tokens (SKILL.md is roughly 4.3k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 7.5k tokens, read only when the agent opens those files.
What are the alternatives to Vllm Rlt Review PR?
Skills that share tags, products or a category with Vllm Rlt Review PR: Find Simplifications (vllm-project/vllm-omni, 7.1k stars), Review PR (vllm-project/vllm-omni, 7.1k stars), SageMaker Serving Image Selection (huggingface/skills, 11k stars) and Hugging Face Local Model Evals (huggingface/skills, 11k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Vllm Rlt Review PR?
ThinkFlowLab (a GitHub organization) maintains it in ThinkFlowLab/vllm-rlt, which has 149 GitHub stars. The repository holds 3 skills in this directory. The repository was last updated on October 11, 2026.
Source: ThinkFlowLab/vllm-rlt on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.