Sketch three directions for this package's icon, show them at 16px and at 256px, and stop before exporting.What comes backThree candidates at both sizes with a one-line rationale each, held for your pick rather than exported.
A favicon, a logo lockup, or a README diagram is work with its own states and its own memory. Felo SuperAgent keeps that work on a persistent canvas, so you check in on it instead of watching it.
cp -r felo-skills/felo-superagent ~/.pi/agent/skills/Pi keeps a deliberately small core and asks you to extend it, so a skill is the intended way to add reach without adding tools to maintain.
The interesting part of a long generation is where it pauses. This one stops when taste is the deciding factor, which is the difference between a tool that helps and one that produces work you throw away.
The package name, the job it does, and the two places the mark has to work: a 16px favicon and a README header.
A geometric monogram, a literal glyph, and a mark built from the shape of the package's own output.
Two directions survived review. The third reads as a different product at small sizes, so it stops rather than spending more effort on it.
SVG and PNG at the sizes the repository actually uses, with a one-line rationale beside each file.
Constraints first, directions second, files last. It is the order that keeps the run short.
The repository needs a mark — a favicon, an icon set, artwork for the README — and there is no designer in the loop.
Name the sizes it has to survive and the two places it will appear. "A modern logo" gives the run nothing to check against, so it optimises for nothing in particular.
Three directions cost less to reject than one polished mark you do not want. The first pass is for choosing, not for shipping.
Look at the candidates at 16px before anything else. Most marks fail there, and it is cheaper to find that out before the export step.
Whoever touches the brand next needs to know why the mark looks like this, and which direction was rejected.
Leaves behindSVG and PNG assets at the sizes the repository uses, plus a short note on the directions that were rejected and the reason.
The bundle is small and the work is not, which is why the root you choose barely matters here. What matters is that the canvas — not the session — is holding the state.
~/.pi/agent/skills/GlobalEvery project on this machine. Put a bundle here once you have read it and want it everywhere.
~/.agents/skills/SharedThe folder other harnesses read too, so one clone serves Pi and whatever else you run.
.pi/skills/ProjectThis repository only, and only after you trust the project. The right place to evaluate a bundle you have not read yet.
pi install npm:@scope/pkgPackageA package's skills/ directory, kept versioned and updatable. The cleaner route once a bundle earns a permanent place.
pi --skill /path/to/felo-superagentBoth hold the job at the reviewable stage instead of letting it run to a finished artifact you may not want.
Sketch three directions for this package's icon, show them at 16px and at 256px, and stop before exporting.What comes backThree candidates at both sizes with a one-line rationale each, held for your pick rather than exported.
The mark is too detailed at small sizes. Reduce it to something that still reads as the same product.What comes backA simplified variant beside the original, so you can see exactly what was dropped and what survived.
Image generation is cheap to start and expensive to review, so the checks are about whether it is respecting your time.
You should see where the run is and what it is waiting for. A job you cannot interrupt is a job you will stop starting.
A run that guesses at your judgement produces work you discard. A blocked state at the right moment is the feature, not a failure.
Check the delivered sizes against what the repository asks for. A mark delivered at one size leaves you doing the resizing anyway.
The artwork is a decision, and decisions get better with something to compare against.
Five skills written for Pi Coding Agent. They follow the same pattern on this platform, so you can combine them in one session without re-explaining your setup.
Start with a skill, then compose the exact models, live data, knowledge, and output tools your Pi workflow needs — the same key every Felo skill already uses.
Use cost-efficient LLM access when your agent needs custom reasoning or generation.
Give agents current web and social signals instead of relying on stale context.
Ground work in reusable knowledge, then turn it into shareable deliverables.
Extend the workflow from research to visual and office-ready results.