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 TianGzlab/OmicsClaw --skill sc-perturbgit clone --depth 1 https://github.com/TianGzlab/OmicsClawWrote 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/tiangzlab/omicsclaw/sc-perturb)<a href="https://agentmods.dev/skills/tiangzlab/omicsclaw/sc-perturb"><img src="https://agentmods.dev/badge/skills/tiangzlab/omicsclaw/sc-perturb/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/tiangzlab/omicsclaw/sc-perturb"><img src="https://agentmods.dev/badge/skills/tiangzlab/omicsclaw/sc-perturb.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Rogue Agent · line 3 Skill modifies its own code, configuration, or behavior at runtime. Self-modification enables an agent to escalate privileges, disable safety constraints, or install persistent backdoors.Fix: Prevent the skill from modifying its own code, SKILL.md, or configuration files. Treat skill files as read-only at runtime.
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 | $0.00084 | $0.01179 |
| Opus 5.5 | $0.00034 | $0.00472 |
| Sonnet 5.5 | $0.00017 | $0.00236 |
| Haiku 4.5 | $0.00008 | $0.00118 |
Grade A, and why
sc-perturb 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 20d 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 — 104 lines — stays where its author put it; the contents beside it link to each section on GitHub.
sc-perturb
Classify observed Perturb-seq cells with pertpy Mixscape. Guide labels and a
non-targeting control group must already exist; otherwise use sc-perturb-prep.
This is not an in-silico knockout predictor.
Key CLI
python skills/singlecell/scrna/sc-perturb/sc_perturb.py --demo --seed 0 --output /tmp/sc_perturb_demo
python skills/singlecell/scrna/sc-perturb/sc_perturb.py --input prepared.h5ad --pert-key perturbation --control NT --split-by replicate --seed 0 --output results/perturb
mixscape (--method mixscape) is the only method. Tune --n-neighbors,
--logfc-threshold, --pval-cutoff and --perturbation-type KO|OE explicitly.
Workflow
Upstream: sc-perturb-prep, or an AnnData with verified screen labels. Validate
labels, supply/compute PCA, build perturbation signatures, then classify cells.
Both pertpy stages receive --seed (API random_state, default 0).
Downstream: sc-de or sc-enrichment after reviewing target-level calls.
Matrix Contract
Uses normalized expression. Count-like X is saved in layers['counts'] and
log-normalized. Existing X_pca is reused; otherwise PCA is computed before the
legacy normalization step. Prefer an upstream PCA on appropriate normalized
features for real screens. The API leaves its input unchanged.
Inputs & Outputs
Input: .h5ad with a perturbation column, control label and optional split key.
The CLI writes processed.h5ad, report.md, result.json,
reproducibility/commands.sh, tables/mixscape_cell_classes.csv,
tables/mixscape_class_counts.csv, tables/mixscape_global_class_counts.csv
and figures/mixscape_global_classes.png. Plot data and their manifest are in
figure_data/; R-enhanced figures are optional.
Gotchas
result.json→data.params.split_byrecords the CLI's resolved split key. A missing CLI split column warns and disables splitting; the API instead raises unless you passsplit_by=None.tables/mixscape_global_class_counts.csvmust contain biological signal before you interpret a run. Exit 0 alone does not establish an effect; the example asserts KO detection in both known-effect groups.tables/mixscape_cell_classes.csvpreserves target-specific and global classes plusmixscape_class_p_ko(or_oe). Posterior probabilities are not experimental validation.- For pertpy 1.0.3, split-based signatures use controls within each split rather than nearest-neighbour selection.
n_neighborsaffects thesplit_by=Nonepath.
What ships with it
9 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.
- _api.py 3.3 KB runs code
- examples/example_step.py 1.6 KB runs code
- references/methodology.md 851 B
- references/output_contract.md 537 B
- references/parameters.md 463 B
- references/r_visualization.md 474 B
- sc_perturb.py 18 KB runs code
- tests/test_perturb_api.py 2.0 KB runs code
- tests/test_sc_perturb_methods.py 1.6 KB runs code
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.
- 20d ago Changed · -16 lines d6080a4a3fac
- 1mo ago First seen · 120 lines · 84 tokens per session scan A 24d9ac4d3868
sc-perturb is a skill published in the GitHub repository TianGzlab/OmicsClaw (161 stars, last pushed 20d ago), licensed Apache-2.0. It adds 84 tokens to every session and 1,179 once invoked, about $0.0003 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 2026-09-03.
Other skills, from other repositories
Virtual Embryo — atlas data + knowledge graph
Query the Virtual Embryo knowledge graph (mouse/human developmental biology: genes, anatomy, Theiler/Carnegie stages, gene expression, diseases, papers) and its 3D atlas catalog (anatomical OPT/light-sheet volumes + 3D spatial- transcriptomics datasets), and visualise those datasets in 3D with the volume3d / spatial3d…
motif-for-claude-science-skill
Prepares, shares, inspects, or explains the standalone Motif for Claude Science artifact, including sequence inventories, alignments, traces, and browser review.
motif
Use Motif to create and open interactive molecular-biology workbenches for DNA, RNA, and protein records, annotations, alignments, traces, results, and construct designs.
motif-for-codex
Use Motif for molecular-biology work with DNA, RNA, and protein records, including sequence exploration, editing, annotation, alignment, construct comparison, trace analysis, cloning design, results, and portable workbench creation.
motif-for-claude-science
Opens bounded sequence data in the connected Motif workbench or creates a self-contained Motif HTML artifact for visual review.
generate_perturbation_hypothesis
Generate candidate perturbation hypotheses (targets, mechanism, expected signature, experiment plan) for a given biological context and phenotype goal.