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 agentmods add skills/microsoft/agent-lightning/releasenpx skills add microsoft/agent-lightning --skill releasegit clone --depth 1 https://github.com/microsoft/agent-lightningWhat 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 | $0.00063 | $0.02661 |
| Opus 5 | $0.00032 | $0.01331 |
| Sonnet 5 | $0.00013 | $0.00532 |
| Haiku 4.5 | $0.00006 | $0.00266 |
Grade A, and why
release 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 yesterday.
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 — 239 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Release Agent Lightning
Merging a pull request does not publish a stable release. Stable publication is
triggered only by pushing a v* tag to the canonical repository; the tagged
commit is what gets tested, built, and uploaded. That same tag push also deploys
versioned documentation and moves the public stable alias, so a release has two
public side effects, not one.
Establish the release state
- Confirm the repository root, clean working tree, current branch, and remotes.
- Resolve the canonical
OWNER/REPO, its default branch, and its permitted merge methods withgh repo view OWNER/REPO --json nameWithOwner,defaultBranchRef,mergeCommitAllowed,rebaseMergeAllowed,squashMergeAllowed. Identify the local remotes for that repository and the contributor fork by their URLs; do not assume particular remote names or merge settings. - Inspect the release contract in:
.github/workflows/pypi-release.yml.github/workflows/docs.yml.github/workflows/tests.ymlscripts/bump_version.shpyproject.tomlagentlightning/__init__.py
- Confirm the canonical default branch is already green before branching from
it. Resolve its current commit with
gh api repos/OWNER/REPO/commits/<default-branch>and inspect that commit's check runs; a general recent-run listing can omit or mix commits. A release branch inherits every failure that main is carrying. - Query the canonical repository's tags and compare them with the versions
published at
https://pypi.org/pypi/agentlightning/json. Confirm the target version exists in neither place, and stop for an explicit release decision when either of these holds:- A tag exists with no matching PyPI version. A published version is immutable, and its tag must never be reused or moved. A tag that never published is a different situation and still needs a human decision, informed by why it did not publish. See "Recovering a tag that never published" below.
- The proposed bump would skip a version that was tagged but never published.
- Treat verified PyPI trusted-publisher configuration for the canonical
repository and
pypi-release.ymlas a prerequisite. If it cannot be inspected directly, require confirmation from an authorized PyPI project owner before pushing the release tag.
What ships with it
1 file 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.
- yesterday First seen · 239 lines · 63 tokens per session scan A 44be3b95bec6
release is a skill published in the GitHub repository microsoft/agent-lightning (17,926 stars, last pushed 4d ago), licensed MIT. It adds 63 tokens to every session and 2,661 once invoked, about $0.0003 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 2026-08-30.
Other skills, from other repositories
AudioCraft
PyTorch library for audio generation including text-to-music (MusicGen) and text-to-sound (AudioGen). Use when you need to generate music from text descriptions, create sound effects, or perform melody-conditioned music generation.
LM Evaluation Harness
Evaluates LLMs across 60+ academic benchmarks (MMLU, HumanEval, GSM8K, TruthfulQA, HellaSwag). Use when benchmarking model quality, comparing models, reporting academic results, or tracking training progress. Industry standard used by EleutherAI, HuggingFace, and major labs. Supports HuggingFace, vLLM, APIs.
Weights & Biases
Track ML experiments with automatic logging, visualize training in real-time, optimize hyperparameters with sweeps, and manage model registry with W&B - collaborative MLOps platform.
llama.cpp
Run LLM inference with llama.cpp on CPU, Apple Silicon, AMD/Intel GPUs, or NVIDIA — plus GGUF model conversion and quantization (2–8 bit with K-quants and imatrix). Covers CLI, Python bindings, OpenAI-compatible server, and Ollama/LM Studio integration. Use for edge deployment, M1/M2/M3/M4 Macs, CUDA-less…
Obliteratus
Remove refusal behaviors from open-weight LLMs using OBLITERATUS — mechanistic interpretability techniques (diff-in-means, SVD, whitened SVD, LEACE, SAE decomposition, etc.) to excise guardrails while preserving reasoning. 9 CLI methods, 28 analysis modules, 116 model presets across 5 compute tiers, tournament…
Segment Anything
Foundation model for image segmentation with zero-shot transfer. Use when you need to segment any object in images using points, boxes, or masks as prompts, or automatically generate all object masks in an image.