Leon is an open-source personal AI assistant that uses tools, memory, context, and agent-like execution to understand goals and complete tasks. People use it as a practical assistant that can run locally and work across structured workflows, while the catalogue skills support its operation.
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 leon-ai/leon --skill codinggit clone --depth 1 https://github.com/leon-ai/leonWrote 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/leon-ai/leon/coding)<a href="https://agentmods.dev/skills/leon-ai/leon/coding"><img src="https://agentmods.dev/badge/skills/leon-ai/leon/coding/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/leon-ai/leon/coding"><img src="https://agentmods.dev/badge/skills/leon-ai/leon/coding.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 | $0.00024 | $0.00900 |
| Opus 5.5 | $0.00010 | $0.00360 |
| Sonnet 5.5 | $0.00005 | $0.00180 |
| Haiku 4.5 | $0.00002 | $0.00090 |
Grade A, and why
coding 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 5d 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 — 38 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Coding
Use Leon's existing agent loop and tools to complete coding tasks directly. Respect the owner's requested scope: analysis is not authorization to edit, commit, push or publish. Delegate only when the owner asks or has an applicable preference.
For requested delegation, use the owner's preferred available agent and interface. Pass the repository, task, applicable instructions and authorized scope. Inspect and verify the actual result; submitting a prompt is not completion. If delegation is unavailable, explain briefly and continue directly within the authorized scope.
Repository and instructions
- Establish the exact workspace and working directory from the request and live observations. Read relevant manifests and nearby implementations; use existing helpers before adding abstractions.
- Before editing, discover and read applicable
AGENTS.mdfiles from the repository root through each target directory. Deeper instructions apply only within their directory and take precedence over broader project rules. For a non-Git workspace, use the owner-selected workspace boundary; do not inherit unrelated parent-directory instructions. - Record the workspace, applicable instruction paths, constraints and acceptance criteria from the request and supplied references in plan details. Pair each criterion with its required check. Before editing a new directory, discover its nested instructions. After compaction or continuation, reload applicable instruction files before further edits; summaries are not substitutes for their full current content.
- Inspect Git status and existing staged/unstaged diffs. Record unrelated changed and untracked files and preserve their content. Do not reset, clean, stage or commit owner work without authorization. Use an isolated worktree when concurrent work or the requested workflow warrants it.
Investigate and change
- Use bounded file reads, ripgrep and available structural search. Trace definitions and callers; check installed dependencies instead of inventing symbols or APIs. Batch independent discovery calls.
- For a bug, reproduce the failure with a relevant check before changing the implementation. Inspect existing tests and extend only meaningful coverage. Do not weaken tests to obtain a pass.
- Prefer
file.patchfor changes to existing text, using the current read's content hash when available. Supply unique exact context and preserve unrelated changes. On stale or ambiguous matches, re-read and revise the patch rather than repeating it or overwriting from a partial view. - Use
file.writefor new files; overwriting an existing file requires its complete current content. Follow the project's formatting and dependency-management instructions. - Use
shell.executeCommandfor finite commands and project checks, always with the observed workspace cwd. Discover commands from project configuration rather than guessing a package manager. - Use shell session functions for dev servers, watchers or interactive programs. Record returned session IDs in the plan, follow output cursors, and verify process status. A started process is not a completed or successful check. Never invent handles, replay a command to recover output, or send secrets through stdin.
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.
- 5d ago Changed · +1 lines 00871bcf5be0
- 9d ago First seen · 37 lines · 24 tokens per session scan A 275480dcb905
coding is a skill published in the GitHub repository leon-ai/leon (17,579 stars, last pushed 4d ago), licensed MIT. It adds 24 tokens to every session and 900 once invoked, about $0.0001 per session on Opus 5.5. 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 2026-10-07.
Other skills, from other repositories
marimo-pair
Work inside the user's live marimo notebook from the code editor: run Python in the same kernel the user does, inspect live notebook state, and commit durable notebook changes through code mode. Use whenever you create, analyze, or improve the user's marimo notebook.
fill-model-descriptions
Curate descriptions and provider-specific reasoning options and token limits in packages/llm-info/data/models.yml using models.dev, OpenRouter, and provider documentation. Use when the user asks to populate model descriptions, enrich the model catalog, or curate descriptions after running pnpm sync-models.
lepton-cli
Operate NVIDIA DGX Cloud Lepton through the globally installed lep CLI. Use when the user asks to inspect or manage Lepton workspaces, endpoints/deployments, dev pods, batch jobs, fine-tuning jobs, Ray clusters, Slurm clusters, Dynamo endpoints, storage, secrets, nodes, ingress, templates, logs, authentication, or…
flyenv-failure-boundaries
Use when implementing or reviewing FlyEnv multi-step operations, batch actions, PID candidate filtering, cleanup, supplementary actions, or changes to throw/reject and terminal result handling.
nanoresearch-experiment
Generate a Python code skeleton from an experiment blueprint.
python
The Vovk.ts Python client — generating a typed Python package from a Vovk API with vovk-python, the py / pySrc templates, TypedDict shapes, JSON + JSON Lines consumption, client-side validation, and PyPI publishing. Use whenever the user wants a typed Python SDK, ongoing Python integration, or PyPI publishing …