Gradient
bergside/awesome-design-skills
Smooth color transitions and gradient-rich surfaces for modern, playful interfaces with visual depth.
A skill your agent uses for any task involving the ledidi library — gradient-based design of minimal edits to categorical sequences (DNA/RNA/protein) so that a frozen oracle model predicts a desired…
Apache-2.0Auto-check passed
Install Ledidi
$ npx skills add jmschrei/ledidi --skill ledidi -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install jmschrei/ledidi ledidi --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/jmschrei/ledidi.git skills-src && mkdir -p .claude/skills && cp -r skills-src/ledidi/_skills/data .claude/skills/ledidi && 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 "ledidi" agent skill from https://github.com/jmschrei/ledidi/tree/master/ledidi/_skills/data into .claude/skills/ledidi/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "ledidi", 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/jmschrei/ledidi/tree/master/ledidi/_skills/dataType 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 jmschrei/ledidi --skill ledidi -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install jmschrei/ledidi ledidi --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/jmschrei/ledidi.git skills-src && mkdir -p .agents/skills && cp -r skills-src/ledidi/_skills/data .agents/skills/ledidi && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "ledidi" agent skill from https://github.com/jmschrei/ledidi/tree/master/ledidi/_skills/data into .agents/skills/ledidi/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "ledidi", 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 jmschrei/ledidi --skill ledidi -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install jmschrei/ledidi ledidi --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/jmschrei/ledidi.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/ledidi/_skills/data .cursor/skills/ledidi && 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 "ledidi" agent skill from https://github.com/jmschrei/ledidi/tree/master/ledidi/_skills/data into .cursor/skills/ledidi/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "ledidi", 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/jmschrei/ledidi.git --path ledidi/_skills/data--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 jmschrei/ledidi --skill ledidi -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install jmschrei/ledidi ledidi --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/jmschrei/ledidi.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/ledidi/_skills/data .gemini/skills/ledidi && 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 "ledidi" agent skill from https://github.com/jmschrei/ledidi/tree/master/ledidi/_skills/data into .gemini/skills/ledidi/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "ledidi", 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 jmschrei/ledidi ledidiInstalls 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 jmschrei/ledidi --skill ledidi -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/jmschrei/ledidi.git skills-src && mkdir -p .github/skills && cp -r skills-src/ledidi/_skills/data .github/skills/ledidi && 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 "ledidi" agent skill from https://github.com/jmschrei/ledidi/tree/master/ledidi/_skills/data into .github/skills/ledidi/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "ledidi", 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 jmschrei/ledidi --skill ledidi -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install jmschrei/ledidi ledidi --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/jmschrei/ledidi.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/ledidi/_skills/data .opencode/skills/ledidi && 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 "ledidi" agent skill from https://github.com/jmschrei/ledidi/tree/master/ledidi/_skills/data into .opencode/skills/ledidi/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "ledidi", 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
ledidi- GitHub stars
- 116
- Token cost
- ~1.8k tokens
- SKILL.md length
- 825 words
- Files
- 20 (incl. references)
- Skills in repo
- 1
- Repo updated
- First seen
- Licence
- Apache-2.0
At a glance
A skill your agent uses for any task involving the ledidi library — gradient-based design of minimal edits to categorical sequences (DNA/RNA/protein) so that a frozen oracle model predicts a desired…
- Editing a sequence
- SKILL.md covers Read these first, Getting the oracle side right, Task → reference file and The rest of the package, plus 1 more section
- Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md
- Knocking out a motif
What it does
Ledidi is an agent skill from jmschrei/ledidi. Use for any task involving the ledidi library — gradient-based design of minimal edits to categorical sequences (DNA/RNA/protein) so that a frozen oracle model predicts a desired output. Triggers on designing or editing a sequence, inserting or knocking out a motif or binding site, hitting a target model output, cell type-specific or output-specific element design, affinity catalogs, in-painting, constraining where edits may be made, pruning edits, or balancing several oracle models in one design. This is a…
Its SKILL.md is about 1.8k tokens, which your agent loads only when the skill is triggered. The skill folder holds 20 other files, including reference files (for example `references/catalogs-and-repeats.md`, `references/custom-losses.md` and `references/designer-object.md`).
The repository describes itself as: Ledidi turns any machine learning model into a biological sequence editor, allowing you to design sequences with desired properties. The licence is Apache-2.0.
When your agent uses it
- Editing a sequence
- Knocking out a motif
- Hitting a target model output
- Cell type-specific
Example prompts
- “/ledidi”
What it can do on your machine
Read from SKILL.md and the folder at commit 673efe4. 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
Ledidi loads about 1.8k tokens when it runs, and up to ~29k if it reads all its reference files. Until then it costs about 157 tokens; SKILL.md has 825 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 jmschrei/ledidi at commit 673efe4, republished under its Apache-2.0 licence (© jmschrei). 825 words, ~1,807 tokens.
Download SKILL.mdSave it as
.claude/skills/ledidi/SKILL.md (or your agent's skills folder). This skill also uses 19 other files; get the full folder from GitHub.- name
- ledidi
- description
- Use for any task involving the ledidi library — gradient-based design of minimal edits to categorical sequences (DNA/RNA/protein) so that a frozen oracle model predicts a desired output. Triggers on designing or editing a sequence, inserting or knocking out a motif or binding site, hitting a target model output, cell type-specific or output-specific element design, affinity catalogs, in-painting, constraining where edits may be made, pruning edits, or balancing several oracle models in one design. This is a router skill — read the relevant file under references/ for details and footguns before writing ledidi code.
ledidi
ledidi inverts the usual training loop: the oracle model is frozen and the
data is optimized. It learns a continuous weight matrix, samples one-hot edits
to a template sequence from a Gumbel-softmax, and pushes those edits until the
oracle predicts what you asked for — while an input loss keeps the number of
edits small. Any differentiable PyTorch model that maps a one-hot sequence to a
prediction becomes a sequence editor.
This skill is a router. Each topic below has a reference file with exact signatures and footguns. Read the relevant reference file before writing code — do not rely on memory of the API. Several of ledidi's failure modes are silent: a plausible-looking argument can broadcast against the wrong shape, protect the region you meant to edit, or prune against the wrong objective, with no exception raised.
tangermeme is a hard dependency of ledidi (it supplies the input validation),
and it owns everything that happens around a design — one-hot encoding, FASTA
and loci I/O, attributions, motif scanning, logo plotting. Install its skill
too (tangermeme-install-skills) and consult it for those steps rather than
reinventing them here.
Is ledidi even the right tool? ledidi is gradient-based and finds small,
targeted edits to an existing template. For discrete design — implanting motifs
from a library, screening random candidates, greedy or beam substitution — use
tangermeme.design instead (screen, greedy_substitution, beam_substitution,
greedy_marginalize); its skill's references/design.md covers them. Note that
tangermeme.design requires a per-candidate loss (reduction='none') to rank
edits, which is the opposite of ledidi's requirement that the output loss
return a scalar — do not carry that habit across.
Read these first
The objective,
references/objective.md— every design isoutput_loss(y_hat, y_bar) + l * input_loss, where the input loss is the mean number of edits per sequence.lis the exchange rate between "hit the target" and "make few edits", and it is the knob you will actually tune. Also covers what theverboselog lines mean and why the returned design is the best-scoring iterate rather than the last one.The oracle contract,
references/oracle-contract.md— what your model must satisfy before any of this works: differentiable,model(X)sliceable as[:, target], frozen and.eval()ed for you. Read it before wrapping anything.
Getting the oracle side right
Most real designs fail here, not in the optimizer. Three distinct situations, each with its own file:
| Situation | Read |
|---|---|
| one multi-task model, you want some of its outputs | references/multi-task-models.md |
| several independent models balanced in one design | references/multiple-models.md |
| the model's input window ≠ the sequence you want to design | references/receptive-field.md |
Task → reference file
| If the task is… | Read |
|---|---|
| starting from scratch — a real oracle, end to end: design → prune → validate → plot | references/pipeline.md |
| a first design / learning the mechanics with no downloads | references/first-design.md |
tensor shapes, dtypes, or a ValueError/TypeError you do not understand | references/io-and-validation.md |
| forbidding edits at certain positions | references/masks.md |
| forbidding or forcing specific characters, or setting soft priors | references/initial-weights.md |
| letting ledidi fill in a blanked region (in-painting) | references/inpainting.md |
a non-MSE objective: MinGap, GapLoss, rewards, one-sided or ballpark losses, profiles | references/custom-losses.md |
| designing against a range of target strengths (affinity catalog), or repeats | references/catalogs-and-repeats.md |
| sampling many designs cheaply, or reusing a fitted designer | references/designer-object.md |
| trimming unnecessary edits after design | references/pruning.md |
| checking a design is real and not oracle exploitation | references/validating-designs.md |
| plotting losses, edit maps, or edits on attribution tracks | references/plotting.md |
| a CUDA out-of-memory error | references/memory-and-oom.md |
| reproducibility, seeding, CPU vs GPU | references/reproducibility.md |
Show full SKILL.md (274 more words)Show less
The rest of the package
ledidi.ledidi— the function you almost always call. Handles device placement, repeats, affinity catalogs, and post-fit sampling.ledidi.Ledidi— the underlyingtorch.nn.Moduleoptimizer (fit_transform,forward). Reach for it only to fit once and sample repeatedly →references/designer-object.md.ledidi.losses.MinGap— output-specific design without target values.ledidi.wrappers.DesignWrapper— concatenate several oracles into one.ledidi.pruning.greedy_pruning— post-hoc edit trimming.ledidi.plot—plot_loss,plot_history,plot_edits.
Conventions
- Tensor layout
(batch, n_channels, length),torch.float32, one-hot along the channel axis (DNA: 4 channels orderedA, C, G, T). The templateXyou pass in has a batch dimension of exactly 1; the returned designs have a batch dimension ofbatch_size. - Naming
Xtemplate,X_bardesigned sequences,y_bardesired output,y_hatpredictions,X_attrattributions. devicedefaults to'cuda', not to "CUDA if available". On a CPU-only machine you must passdevice='cpu'explicitly or the call raises.ledidi()moves the model in place, but not your template. Designs come back on the device while yourXstays where it was, so.to(device)yourXyourself — otherwise pruning, edit diffs, anddesigner(X)all raiseRuntimeError: Expected all tensors to be on the same device. This is the most common first error.- Substitutions only. ledidi changes characters in place; it never inserts or deletes, so the length is fixed for the whole design.
- ledidi's keywords are not tangermeme's.
ledidi()forwards**kwargstoLedidi.__init__, which takes no**kwargs, so a tangermeme habit likeoutput_mask=,args=, orfunc=raisesTypeErrorrather than being silently ignored. Output selection istarget→references/multi-task-models.md. - Designs from the default return path carry an autograd graph (
n_samplesdraws are detached)..detach()before holding many, or before handing them to code that assumes plain tensors →references/memory-and-oom.md.
© jmschrei, 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 19 other files (references) in ledidi/_skills/data of jmschrei/ledidi.
- SKILL.md
- references/catalogs-and-repeats.md
- references/custom-losses.md
- references/designer-object.md
- references/first-design.md
- references/initial-weights.md
- references/inpainting.md
- references/io-and-validation.md
- references/masks.md
- references/memory-and-oom.md
- references/multi-task-models.md
- references/multiple-models.md
- references/objective.md
- references/oracle-contract.md
- references/pipeline.md
- references/plotting.md
- references/pruning.md
- references/receptive-field.md
- references/reproducibility.md
- references/validating-designs.md
Open the folder on GitHubat commit 673efe4
Compare with similar skills
Ledidi 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 |
|---|---|---|---|---|---|---|
| Ledidi this skilljmschrei/ledidi | 116 | — | ~1.8k | Automated safety check: Pass | Apache-2.0 | |
| Gradientbergside/awesome-design-skills | 3.1k | 1 repos | ~868 | Automated safety check: Pass | MIT | |
| Gradient Designsickn33/agentic-awesome-skills | 47k | 1 repos | ~3k | Automated safety check: Pass | MIT | |
| Gradient Methodsparcadei/Continuous-Claude-v3 | 3.9k | 2 repos | ~1k | Automated safety check: Notes | MIT | |
| HTML Ppt Obsidian Claude Gradientnexu-io/open-design | 100k | — | ~1.4k | Automated safety check: Pass | Apache-2.0 | |
| Warp Debug GradientsNVIDIA/skills | 3.6k | — | ~4.8k | Automated safety check: Pass | Apache-2.0 |
Similar skills
Gradient Design
sickn33/agentic-awesome-skills
Web and App implementation guide for Gradient Design. An agent skill from sickn33/agentic-awesome-skills.
Gradient Methods
parcadei/Continuous-Claude-v3
Problem-solving strategies for gradient methods in optimization
HTML Ppt Obsidian Claude Gradient
nexu-io/open-design
OpenDesign's enterprise AI-adoption brief: local-first agents at work, the risk controls, the ROI, and the rollout plan.
- Official
Warp Debug Gradients
NVIDIA/skills
A skill your agent uses to diagnose and fix incorrect gradients in differentiable Warp programs.
Gradient Clipping Helper
jeremylongshore/tons-of-skills-marketplace
Configure with gradient clipping helper operations. An agent skill from jeremylongshore/tons-of-skills-marketplace.
Questions about Ledidi
What does Ledidi do?
A skill your agent uses for any task involving the ledidi library — gradient-based design of minimal edits to categorical sequences (DNA/RNA/protein) so that a frozen oracle model predicts a desired…. Ledidi is an agent skill from jmschrei/ledidi. Use for any task involving the ledidi library — gradient-based design of minimal edits to categorical sequences (DNA/RNA/protein) so that a frozen oracle model predicts a desired output.
When should I use Ledidi?
Ledidi fits situations like: editing a sequence; knocking out a motif; hitting a target model output; cell type-specific.
How do I install Ledidi in Claude Code?
Run `npx skills add jmschrei/ledidi --skill ledidi -a claude-code`. Or copy the skill folder (ledidi/_skills/data in jmschrei/ledidi) into .claude/skills/ledidi in your project. Claude Code loads it when a task matches its description.
How do I install Ledidi in Codex?
Run `npx skills add jmschrei/ledidi --skill ledidi -a codex`. Or copy the skill folder (ledidi/_skills/data in jmschrei/ledidi) into .agents/skills/ledidi in your project. Codex loads it when a task matches its description.
Can I use Ledidi 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 jmschrei/ledidi --skill ledidi -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/ledidi, .gemini/skills/ledidi, .github/skills/ledidi and .opencode/skills/ledidi in your project.
What does Ledidi need to run?
SKILL.md names no scripts, command-line tools or credentials: Ledidi is instructions for the agent only.
Does Ledidi 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 Ledidi 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 Ledidi use?
Ledidi 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 Ledidi use?
About 1.8k tokens (SKILL.md is roughly 7.2k 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 27k tokens, read only when the agent opens those files.
What are the alternatives to Ledidi?
Skills that share tags, products or a category with Ledidi: Gradient (bergside/awesome-design-skills, 3.1k stars), Gradient Design (sickn33/agentic-awesome-skills, 47k stars), Gradient Methods (parcadei/Continuous-Claude-v3, 3.9k stars) and HTML Ppt Obsidian Claude Gradient (nexu-io/open-design, 100k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Ledidi?
jmschrei (a GitHub user) maintains it in jmschrei/ledidi, which has 116 GitHub stars. The repository was last updated on September 16, 2026.
Source: jmschrei/ledidi on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.