3.x renamed the streaming helper. Every wrapper importing it by the old name breaks without a warning.
Breaking changeCheck whether anyone else is hitting the same wall
A stack trace stops being unique the moment other people upgrade. Felo X Search reads what is being posted about a version, an API, or an outage and returns the posts, the time, and the signal each one carries.
cp -r felo-skills/felo-x-search ~/.pi/agent/skills/
- Bundle
- felo-skills/felo-x-search
- Query
- Keyword, handle, or topic
- Comes back as
- Posts with a signal tag
- Most useful
- Release week
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.
- Loads from
- ~/.pi/agent/skills/ · .pi/skills/ · packages
- Install is
- A folder copy, or pi install npm:@scope/pkg
- Worth knowing
- No sandbox — read the SKILL.md before you mount it
A read-only bundle, loaded the same way as the rest
Nothing here can post, so the security question is smaller than usual — but it still makes network calls, and Pi still does not sandbox anything. Read it, then decide how wide the root should be.
- 01
~/.pi/agent/skills/GlobalEvery project on this machine. Put a bundle here once you have read it and want it everywhere.
- 02
~/.agents/skills/SharedThe folder other harnesses read too, so one clone serves Pi and whatever else you run.
- 03
.pi/skills/ProjectThis repository only, and only after you trust the project. The right place to evaluate a bundle you have not read yet.
- 04
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-x-searchPosts sorted into signals, so a lead is not mistaken for evidence
The sort is the useful part. It separates a reproducible report from a complaint, and a maintainer's answer from speculation, before you spend an afternoon on the wrong thread.
Same timeout here after the upgrade. Pinning back to 2.4.1 made it disappear.
ReproducibleThe docs still show the 2.x signature on the main page. Filed an issue with the diff.
Docs lagNo problems on our side, but we never used the streaming path, so that may not mean much.
Scope limitThe signal tag is a claim about the post, so it is worth checking one or two against the thread before you act on the set.
Three things a post cannot tell you
Social results are fast and shallow by nature. The limits are worth stating before a release decision rests on them.
A post is a lead, not evidence
One reproducible report outweighs a hundred complaints. The signal tag exists so you can tell those apart before you change a version.
Volume is not severity
A loud thread about a cosmetic regression can bury a quiet one about data loss. Sort by what the post claims, not by how many people repeated it.
The API has a rate budget
Set that against how often your routine runs. A watcher polling every minute spends the day rate-limited and reports nothing.
Two checks worth running before you ship the upgrade
Ask for the reproducibility ranking, not for a mood. The first question decides whether you wait.
Search X for people reporting timeouts after the 3.x upgrade, and rank what you find by how reproducible it looks.What comes backPosts grouped by signal, reproducible reports first, speculation kept visibly separate.
Is anyone from the vendor replying to this issue on X? Quote them with dates.What comes backReplies from named maintainers with timestamps — or a clear statement that nobody official has answered yet.
Three checks on a social result set
Fabrication here looks like a confident summary, so verify the pieces rather than the tone.
Handles resolve
Open one handle on X directly. An account that does not exist is the fastest way to catch an invented result set.
Times fit the version
Posts should be recent relative to the release you are asking about, not merely recent relative to today.
The tag matches the post
If a single unverified complaint is labelled reproducible, the summarising step is doing more than the evidence supports.
What Pi users ask about Felo X Search
Can it read a whole thread?
Can it post or reply?
How current are the results?
What is it bad at?
Give the session a way to check the outside world
One bundle, one key, four roots to choose from. Load it for a single session first and see whether the answers hold up.
Every Felo skill for Pi Coding Agent
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.
One API key. A full agent toolkit.
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.
Models
Use cost-efficient LLM access when your agent needs custom reasoning or generation.
- LLM API
- Chat completions
- Model access
Live information
Give agents current web and social signals instead of relying on stale context.
- Web Search
- X Search
- Web Fetch
Knowledge and output
Ground work in reusable knowledge, then turn it into shareable deliverables.
- LiveDoc
- PPT generation
- Mindmaps
Creative tools
Extend the workflow from research to visual and office-ready results.
- Image generation
- Page tools
- Office outputs