agentleFS
cwhited26cwhited26/pocket-agent-starter-skills/skills/research/source-verification/SKILL.md

source-verification

Skill0 starsChanged 4 months agoMITView on GitHub

Attach the URL or source to every factual claim in research, and flag anything you couldn't verify instead of stating it as fact. Any research deliverable the owner will act on or repeat. Especially numbers, prices, and competitive claims that could be wrong.

5 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every number, price, and named fact gets its source inline — the URL, the page, the date pulled.

    line 20
  2. Distinctive:

    If you can't find the source, the claim doesn't ship as fact.

    line 21
  3. Distinctive:

    When two sources disagree, say so and show both — don't silently pick one and present it as settled.

    line 24

“Common” is measured: kept by at least 6 in 10 of the 5 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Every Claim Points at Its Source

A research claim without a source is a guess wearing a suit. Point every fact at where it came from, and when you can't verify something, say so plainly rather than laundering it into certainty.

The technique

  • Every number, price, and named fact gets its source inline — the URL, the page, the date pulled.
  • If you can't find the source, the claim doesn't ship as fact. Flag it: "unverified — couldn't confirm."
  • Separate what you found from what you inferred. An inference is fine; passing it off as a sourced fact isn't.
  • Date the pull. Prices and pages change; a fact true last month may be stale, and the date tells the reader how much to trust it.
  • When two sources disagree, say so and show both — don't silently pick one and present it as settled.

Do this, not that

Do: "Their entry price is $79/mo (their /pricing page, pulled 2026-06-09). Their churn is reportedly high — unverified, that's from one forum thread, treat as a rumor, not a number."

Don't: State "they charge $79 and have terrible retention" as flat fact, where the price is sourced but the retention is a stranger's comment you can't back up — now the owner repeats a rumor as data.

Why this works

Research gets repeated. The owner will say it on a call, put it in a deck, price against it. If a claim is wrong and there's no source, nobody catches it until it's already cost something. Sourcing every fact makes the work checkable and the mistakes findable, and flagging the unverified keeps a guess from hardening into a "fact" that wasn't.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.