00 Research skill
Free Claude research skill: Research that says “I don’t know.”
AI research reads finished, even where it guessed. This skill stamps every line it writes: seen in a source, worked out, disputed or not found. What it can’t find, it writes down as a gap.
One skill · four guides · one check script · free, MIT License
research--
12 OCT 2026
--- source_count: 34 confidence_distribution: observed: 172, inferred: 30, contested: 1, unknown: 17 coverage: 10 of 10 sub-questions resolved ---
## Key findings
1. [OBSERVED] YouTube Help's "Add video thumbnails" page recommends a custom thumbnail resolution of 3840 x 2160 pixels for videos. [S1] (as of 2026-10-12, page undated) (single-source)
11. [INFERRED] A video thumbnail file larger than 2 MB and up to 50MB can be uploaded only from desktop, because YouTube Help caps mobile video thumbnail uploads at 2 MB. Rests on findings 9, 10. [S1]
## Open questions and gaps
7. [UNKNOWN] Any peer-reviewed study measuring the effect of faces or facial expressions in YouTube thumbnails on click-through rate, views or watch time. Blocking reason: none surfaced in the searches run; …
use a face? no study found.
the file says so.
01 Confidently wrong
AI called my strategy bulletproof.
It missed the main competitor, which already did almost everything the product did. The answer read finished. Nothing in it said where it was guessing.
I believed it six times this year
-
#1 · The prompt tool
Nobody has built this.
false
-
#2 · The dog collar
Nobody has built this.
false
-
#3 · name withheld
Nobody has built this.
false
-
#4 · name withheld
Nobody has built this.
false
-
#5 · name withheld
Nobody has built this.
false
-
#6 · name withheld
Nobody has built this.
false
“I believed ‘nobody has built this’ six times this year. Six times it was false.”
Two of the six, the prompt tool and the dog collar, are in video 01. The other four stay unnamed.
Receipt · AI search, tested 6 Mar 2025
60%+ wrong
8 AI search tools, 1,600 citation queries. The premium versions got more wrong than the free ones.
Source Tow Center for Digital Journalism, Columbia Journalism Review cjr.org (opens in a new tab)Receipt · Same test, ChatGPT searchReceipt · ChatGPT search 6 Mar 2025
“signaled a lack of confidence just fifteen times out of its two hundred responses”
it doubted itself 15 times in 200.
mine has to stamp every line.
02 Four stamps
Every line gets one of four stamps.
So you, and the next AI that reads the file, know how far to trust each line. These are the counts from my 12 YouTube growth files.
A3 · four stamps, one per line
-
2,237
81.0% of lines
A source it opened says so. The only stamp that reads as fact.
-
287
10.4% of lines
Worked out from its own findings, with the finding numbers cited.
-
30
1.1% of lines
Credible sources disagree. Each side stays in, with why they differ.
no quiet winner
-
209
7.6% of lines
Nothing credible found. Written down as a gap, never filled with a guess.
the honest one.
a short honest file beats
a long fabricated one.
Counted from the tags in my 12 files: observed 2,237 · inferred 287 · contested 30 · unknown 209. Total 2,763.
04 The rules
Five things it never does.
Each one is a way AI research looks finished when it isn’t. The skill would rather hand you a shorter file.
Rules · they override looking complete
From SKILL.md
-
01 Never
Cite a source it didn’t open.
No remembered links, no rebuilt URLs. If it didn’t open it this run, it doesn’t exist.
-
02 Never
State a fact it can’t trace.
No source means UNKNOWN, written down as a gap.
-
03 Never
Invent precision.
Numbers, units and dates exactly as the source states them. No extra decimals.
-
04 Never
Promote a guess.
Passing INFERRED off as OBSERVED is the one fatal error.
-
05 Never
Pad the file to look finished.
If the core can’t be sourced, it stops and tells you to narrow the question.
Three passes, then it stops
- 01
Break it down, search wide.
Each sub-question searched on its own, from more than one side, including the side that would prove it wrong.
- 02
Fill the gaps.
Weak spots and disagreements get a second, targeted pass.
- 03
Try to break it. Then ship.
Every load-bearing claim gets a query built to disprove it. What doesn’t survive is cut.
pass 3 is the one that stings.
05 Receipts
I ran it on my own channel first.
Before the first video, I asked how YouTube growth actually works. Twelve questions, twelve files, every line stamped. My launch plan cites them.
The drawer · 12 files, compiled 12 Oct 2026
-
01
Algorithm and discovery
-
02
Analytics and benchmarks
-
03
Channel page and branding
-
04
Descriptions and metadata
-
05
Engagement and promotion
-
06
Monetization and policies
-
07
Retention, hooks and length
-
08
Search and SEO
-
09
Shorts and long-form
-
10
Thumbnails
-
11
Titles
-
12
Upload cadence and timing
this one set my launch day. see below.
- Sources opened
- 616
- 367 tier 1 · 139 tier 2 · 110 tier 3
- Observed
- 2,237
- Inferred
- 287
- Contested
- 30
- Unknown
- 209
- Written down as gaps. Not guessed.
- Stamped lines
- 2,763
- Sub-questions
- 107 of 108
The one left open says so in its file
6. [OBSERVED] YouTube sends a viewer a maximum of 3 upload and live stream notifications from each channel in a 24-hour period. [S2, S1] (as of 2026-10-12, pages undated)
Three videos go public at once. Nothing else that day: no Short, no trailer.
one line, one decision. that’s what the stamps are for.
06 Get the skill
Want research you can check?
One skill, four guides and the check it runs on its own file, in one zip. Read every file before you run it. I would.
- research-skill.zip
- inspected-by-m-deep-research/
- LICENSE : MIT, word for word
- README.md : how to install it
- SKILL.md : the rules and the loop
- references/ : 4 guides
- search-tactics.md : how it searches
- source-directory.md : where good sources live
- output-spec.md : what the file must hold
- verification.md : how it attacks its own claims
- assets/template.md : the file’s skeleton
- scripts/validate_research.py : the check
- Works in
- Claude Code, in the Claude desktop app, with web search on.
- Needs
- Python 3 for the check script. Nothing else.
- Yours to
- Copy, change and share, under the MIT License.
Check your inbox.
- Sent to
- the address you entered
- Skill
- Research skillCheck script
If the download didn’t start, get the file here.
Download the self-checkOne more thing
Want the newsletter to fit what you read? Say yes and I’ll email you a link. Open it and confirm, and a cookie in that browser lets the site keep a list of the pages you read here, with dates, next to your email, for 13 months. I use the list only to choose what I send you, and I share it with no one. Switch it off any time.
Free to copy, change and share, under the MIT License. It comes with no warranty: read a skill before you run it.
How I use your emailBox not ticked? You get the skill and nothing else, and your address is deleted after 30 days.
07 Work with me
Research done. Now build it right.
I build MVPs, do rebuilds and run fix sprints after an inspection. Quoted on a 20-minute call. I do 80 to 90% of the work with my AI skills. My team polishes and tests.
Not ready? Watch the inspections on YouTubeNot ready? Watch on YouTube (opens in a new tab)I read every request myself.
Free, 20 minutes, no code access. For apps with paying users.