Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/PerforatedAI/PerforatedAInpx agentmods add skills/perforatedai/perforatedai/perforatedai-plotWrote 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/perforatedai/perforatedai/perforatedai-plot)<a href="https://agentmods.dev/skills/perforatedai/perforatedai/perforatedai-plot"><img src="https://agentmods.dev/badge/skills/perforatedai/perforatedai/perforatedai-plot/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/perforatedai/perforatedai/perforatedai-plot"><img src="https://agentmods.dev/badge/skills/perforatedai/perforatedai/perforatedai-plot.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 | $0.00090 | $0.01724 |
| Opus 5.5 | $0.00036 | $0.00690 |
| Sonnet 5.5 | $0.00018 | $0.00345 |
| Haiku 4.5 | $0.00009 | $0.00172 |
Grade A, and why
perforatedai-plot 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 26d 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 — 165 lines — stays where its author put it; the contents beside it link to each section on GitHub.
PerforatedAI Plot Skill
Overview
This skill produces one publication-style figure: parameter count on the
x axis, a score on the y axis, one line or scatter stream per model or
configuration. It reads a JSON spec of literal numbers and renders it with
data_processing/plot_streams.py. The spec is the frozen record of the
figure, so every number in the PNG can be traced back.
When to use this skill:
- After a sweep, to draw the best score per dendrite count per model
- After one or more PAI training runs, to draw score against parameters
- When the user hands you numbers and wants them in the shared PAI style
Not for: interpreting results or recommending settings. That is the
perforatedai-analyze skill.
Scripts
All scripts live in data_processing/ of the PerforatedAI repository.
Run them from that directory or by absolute path.
| Script | Purpose |
|---|---|
plot_streams.py |
Render a spec to a PNG |
spec_from_csv.py |
Build a spec from a sweep CSV or from PAI run folders |
pai_style.py |
Shared colors and axes style, imported by the others |
example_spec.json |
A working spec showing every optional field |
Plotting is read-only work and runs on the local machine.
Entry Point
Ask the user one question at a time. Offer a recommended default with each question and wait for the answer before the next one.
Step 0: Choose the source
Ask which of the three sources the figure comes from:
- Sweep CSV: a by-dendrite-separate CSV from
get_wandb_results.py --mode by-dendrite-separate. One stream per model, one point per dendrite count. - PAI run folders: folders holding
<save_name>_best_arch_scores.csvfromUPA.perforate_model(save_name=...). Each stream is a list of folders, one point per folder by default. - Numbers: values the user supplies, or that you pull from W&B in session and write into the spec as literals.
Step 1: Output name and location
Ask for the PNG stem, for example yolo26_cityscapes. Then ask where the
PNG and spec copy should go. Recommend the directory that will hold the
spec. Pass the answer to plot_streams.py as an absolute --out-dir.
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.
- 26d ago First seen · 165 lines · 90 tokens per session scan A 91104551806a
perforatedai-plot is a skill published in the GitHub repository PerforatedAI/PerforatedAI (237 stars, last pushed 2d ago), licensed Apache-2.0. It adds 90 tokens to every session and 1,724 once invoked, about $0.0004 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 2026-09-20.
Other skills, from other repositories
physicsnemo-discover
Official NVIDIA-authored guidance for navigating PhysicsNeMo — pick the model, datapipe, or example for a SciML/AI4Science task (surrogates, forecasting, downscaling, physics-informed, inverse, generative). Points at existing files via live repo search; never writes code. Do NOT use for installation or environment…
HomeSafe-Bench
VLM indoor safety hazard detection benchmark inspired by HomeSafeBench (arXiv 2509.23690).
depth-estimation
Real-time depth map privacy transforms using Depth Anything v2 (CoreML + PyTorch).
ml-for-aec
Computer vision for buildings, image-to-floorplan, generative ML models, performance prediction, structural analysis ML, energy prediction, natural language to design, and point cloud ML for AEC computational design.
ml-architecture-diagram
Create accurate, publication-ready, editable machine-learning and deep-learning architecture diagrams from model code, configuration files, model summaries, graph exports, or written specifications. Use for neural-network figures, architecture schematics, model block diagrams, training/inference diagrams, paper…
depth-estimation
Real-time depth map privacy transforms using Depth Anything v2 (CoreML + PyTorch).