Pi ships a deliberately small agent core and asks you to extend it with extensions, skills, prompt templates, themes, and packages. Felo skills extend it without adding code to the core — live search, page extraction, transcripts, and deck output arrive as plain Markdown bundles.
Minimal terminal coding harness
pi # then ask for a skill that fetches a page as MarkdownPi's documentation is direct about scope: the core is intentionally small, and features such as sub-agents and plan mode are left out. Skills are the intended extension point for task knowledge — including tasks that leave the terminal.
Pi loads skills from ~/.pi/agent/skills, ~/.agents/skills, project-level .pi/skills and .agents/skills, or from a package's skills/ directory. Drop in what you need; leave the rest out.
Pi stays minimal on purpose, so a skill is how you add reach without adding tools to babysit. Results arrive dated and linked, ready to cite in the same reply.
Find the recommended Node version for this dependency and cite where you read it.cp -r felo-skills/felo-search ~/.pi/agent/skills/Reviewers read decks, not terminal scrollback. This turns what the session found into slides without leaving Pi's workflow.
Summarize this refactor in five slides: what moved, what broke, what needs review.cp -r felo-skills/felo-slides ~/.pi/agent/skills/Pi has no browser pane, so video content is normally out of reach. This pulls captions and hands back text the session can grep.
Get the transcript of this talk and pull out every benchmark number it quotes.cp -r felo-skills/felo-youtube-subtitling ~/.pi/agent/skills/Worth running before a release: check whether other people on this version are hitting the same error you are staring at.
Search X for errors from this library version and list the three most repeated ones.cp -r felo-skills/felo-x-search ~/.pi/agent/skills/When the project needs a favicon, a logo lockup, or a product shot, the same session can generate and iterate on it instead of handing you a prompt to run elsewhere.
Sketch a logo for this package in three directions and save the variants as files.cp -r felo-skills/felo-superagent ~/.pi/agent/skills/Pi loads skills from a folder you own, so adding one is a copy. Pick the job you want handled next: research, extraction, slides, transcripts, social monitoring, or a long-running workflow.
Add on Pi
Live web research with sources you can cite
cp -r felo-skills/felo-search ~/.pi/agent/skills/Turn pages into structured, reusable data
cp -r felo-skills/felo-web-fetch ~/.pi/agent/skills/Answers grounded in your own documents
cp -r felo-skills/felo-livedoc ~/.pi/agent/skills/Transcripts from talks, demos, and reviews
cp -r felo-skills/felo-youtube-subtitling ~/.pi/agent/skills/Track posts and replies on X
cp -r felo-skills/felo-x-search ~/.pi/agent/skills/Generate a deck from a single prompt
cp -r felo-skills/felo-slides ~/.pi/agent/skills/Turn notes and URLs into a deck outline
cp -r felo-skills/felo-content-to-slides ~/.pi/agent/skills/Long-running workflows that keep their memory
cp -r felo-skills/felo-superagent ~/.pi/agent/skills/Capture pages as snapshots and Markdown
cp -r felo-skills/doc-screenshot-agent ~/.pi/agent/skills/Draft X posts from your own material
cp -r felo-skills/felo-twitter-writer ~/.pi/agent/skills/Generate a landing page from a brief
cp -r felo-skills/felo-landingpage ~/.pi/agent/skills/Apple product research before you buy
cp -r felo-skills/apple-buy-advisor ~/.pi/agent/skills/Project-scoped instead? Copy into .pi/skills/ and trust the project when Pi prompts.
Pi users usually install one skill folder first, keep the session open, and only add the output skills once research works.
Pi reads skills from the user layer, the project, or a package, so start with the narrowest scope.
The value shows up the first time the session answers with a source instead of a guess.
Once research works, the same session can produce what the team actually reads.
Two paths: the direct copy works everywhere, and Pi packages are the cleaner option once you want the same bundle on several machines.
Use the npm install or the shell installer, then run pi inside a project directory and authenticate with /login or a provider API key.
npm install -g --ignore-scripts @earendil-works/pi-coding-agent
# or
curl -fsSL https://pi.dev/install.sh | sh
# then, in a project directory
piSkills call the Felo CLI, so install it once and store the key outside the repository.
npm install -g felo-ai
felo config set FELO_API_KEY your-api-key-hereGlobal skills live under ~/.pi/agent/skills. Project skills belong in .pi/skills and only load after you trust the project. If you keep one shared folder for several harnesses, ~/.agents/skills works too.
git clone https://github.com/Felo-Inc/felo-skills.git
# Every project on this machine
mkdir -p ~/.pi/agent/skills
cp -r felo-skills/felo-search ~/.pi/agent/skills/
# Or just this repository
mkdir -p .pi/skills
cp -r felo-skills/felo-search .pi/skills/You can also load a bundle without copying it: pi --skill /path/to/felo-search. The flag is additive even when you start with --no-skills.
Pi validates skills against the Agent Skills standard and reports most violations. Ask for the skill list, then run the task you actually care about.
Which skills are available in this session? Then use felo-search
to pull the release notes for our database driver.Three everyday moments where a skill beats another extension.
The stack trace points into a dependency, and your installed version is two releases behind the documentation.
Fetch the current docs for this method, compare the signature with the pinned version, and patch the call site.A patch plus the exact documentation it was based on, instead of a guess that compiles but behaves oddly.
You need to explain a build-versus-buy decision to a team that is not reading the repository.
Research the current pricing and limits for these three services and draft six slides I can edit.A sourced comparison and a slide draft, produced in the same terminal session.
A vendor's launch talk mentions a behavior change that never made it into the changelog.
Pull the transcript, list every behavior change mentioned, and check which ones affect our config.A checklist grounded in the recording, with timestamps you can verify.
Copy one folder into ~/.pi/agent/skills, restart the session, and ask Pi something that needs the live web.
Felo AI is an independent skill publisher. Pi is maintained by Earendil Inc.
Start with a skill, then compose the exact models, live data, knowledge, and output tools your Pi Coding Agent 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.