We moved off that approach in March. The maintenance cost was fine, the review burden was not.
Contradicts the draftFelo X Search on QoderWork
QoderWork
QoderWorkBefore the draft goes out, check what people stopped believing
A document can be internally consistent and still assert something the market moved past. Felo X Search reads what is being posted about a claim and returns the posts with the signal each one carries, so the review is about specific sentences rather than a general impression.
- Install
- Ask in the conversation
- Query
- Claim, product, or handle
- Comes back as
- Posts with a signal tag
- Use
- Pre-publication review
The install request
Please download https://github.com/Felo-Inc/felo-skills/tree/main/felo-x-search
and put it in ~/.qoderwork/skills/, then confirm the skill is loaded.QoderWork runs on your machine and edits local docx and pptx files. A Felo skill supplies the half a local file cannot: current sources, transcripts, and assets that land beside your document.
- Platforms
- macOS 14+, Windows 10+ (64-bit)
- Skills live in
- ~/.qoderwork/skills/
- Install by
- Asking in the conversation
Posts grouped by what they claim, checked against your sentences
The point is not a mood read, it is a line-by-line check. Each claim in the draft gets the posts that support it, the ones that contradict it, and a signal tag saying which kind of evidence each one is.
Same numbers, different conclusion. It depends whether you count abandoned setups in the total.
Method disagreementNobody in our group has tried it at that scale, so I have nothing useful to add.
Scope limitSharing the vendor's thread without comment.
Echo, not reportingAsk for the posts that contradict the draft specifically. A search that only returns agreement is a search that was asked the wrong question.
Three ways a claim check goes soft
All three are quiet, which is why they are worth reading before the first review pass.
A post is a lead, not proof
One practitioner reporting the opposite outcome is a reason to look, not a reason to rewrite. The signal tag exists so the difference survives into your notes.
Agreement is easy to find
Ask for contradicting posts explicitly. A set that only supports the draft was probably asked a question that could only confirm it.
Recency changes the answer
A method critique from last month may already be fixed. The date on each post is what tells you whether the objection still stands.
Two requests that make the review concrete
Both point at a document and a section, which is what keeps a social search from becoming a general impression.
Check the market claims in section two of ~/Documents/launch-brief.md against posts from the last two weeks. Flag any that are disputed.What comes backA short list of claims with supporting and contradicting posts, so the review is about sentences rather than vibes.
Is there anyone reporting the same result we claim, with numbers? Quote them with dates.What comes backQuoteable posts from named accounts, or an honest statement that nobody published figures either way.
Add it on QoderWorkInstall it into the folder the app already watches
Nothing here needs a browser or a login on your machine. The skill reads; the app places the folder; the check runs when a document asks for it.
You
Please download https://github.com/Felo-Inc/felo-skills/tree/main/felo-x-search and put it in ~/.qoderwork/skills/, then confirm the skill is loaded.
QoderWork
Cloned the repository and placed felo-x-search in ~/.qoderwork/skills/. The skill reads posts and returns them with a classification; it cannot post or reply.
You
Read ~/Documents/launch-brief.md and check the three market claims in section two against what people are saying on X this week.
QoderWork
Two claims still hold. The third is disputed on method, with one practitioner reporting the opposite outcome. Sources and dates are attached.
What ends up on disk
~/.qoderwork/skills/felo-x-search/Read-only by design. The SKILL.md lists what it queries and confirms it never writes to an account.~/.qoderwork/skills/The root it matches from. QoderWork picks a skill by its description, so a short list makes the matching predictable rather than clever.
Then ask it to account for itself
Which skill checked the claims, and which posts disagreed with the draft?Three places the check should leave a trace
A verification that lives only in a chat is a verification nobody can audit later.
The draft
Inline noteMark the disputed sentence where it sits, so the person editing sees the objection in context rather than in a separate message.
Source appendix
Link listAdd the posts that shaped the wording. A claim softened to survive review should say what it was softened for.
Review thread
Quoted postsWhen the team argues about a claim, the quoted posts end it faster than any summary of the argument.
Three checks before you act on a claim review
The failure mode here is a confident summary, so verify the pieces rather than the tone.
Open one post
Find a quoted handle on X directly. A handle that does not resolve means the set was assembled rather than collected.
Check the dates against the claim
The post should be recent relative to the practice you are describing, not just recent relative to today.
Look for what was excluded
Ask which posts were left out of the summary and why. An answer that cannot name any is an answer that did not filter.
What QoderWork users ask about Felo X Search
Can it read replies to a specific post?
Yes. When the objection is in the replies rather than the original post, ask for the thread and the answers underneath it.
Can it post our response?
No. It reads. Nothing about it writes to an account, which is what makes it safe on a machine where you keep drafts.
Does this need my X login?
No. The skill reaches an index through the Felo CLI, so your own account is not involved and nothing appears in your timeline.
What is it bad at?
Anything behind a link — full articles, gated research. It reports what was posted, and a fetch is a different job.
Every Felo skill for QoderWork
Five skills written for QoderWork. They follow the same pattern on this platform, so you can combine them in one session without re-explaining your setup.
Felo Search
Felo Slides
Felo YouTube Subtitling
Felo SuperAgentOne API key. A full agent toolkit.
Start with a skill, then compose the exact models, live data, knowledge, and output tools your QoderWork 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