handshake
Skill3 starsChanged 12 days agoMITView on GitHub
Turns the founder folder into the thing that gets one real person to say yes to the next step: a one-page pilot offer for a customer, an NSF Project Pitch or grant answer for a funder, an investor or lender one-pager, a business case for a boss, or an invite for the people who will use it. Always writes the one-sentence, three-sentence and three-minute versions first. Builds only from claims in the evidence ledger that the audience may see, never from simulated buyers, and checks the finished pitch with the ledger before it goes out. Use when the user says "pitch this", "write a pilot offer", "one-pager", "grant pitch", "NSF pitch", "deck for", "email this customer", "business case", or "what do I send them".
5 other projects use a version of this.
Adopted by
Most recently taken up by
- iblai/higher-education-agentstheir own version, first seen Nov 8, 2026
- ZhouYinLong-lab/Dog-Skillstheir own version, first seen Nov 8, 2026
- joansebgarcia/skillstheir own version, first seen Nov 1, 2026
- sorawit-w/agent-skillstheir own version, first seen Nov 1, 2026
- HouseAlone/handshake-skilltheir own version, first seen Sep 24, 2026
Other versions5 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 5
- iblai/higher-education-agents★ 14same name
- ZhouYinLong-lab/Dog-Skills★ 7same name
- sorawit-w/agent-skills★ 2same name
- HouseAlone/handshake-skill★ 0same name
- joansebgarcia/skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository13 other files it gives its agents.
- burn-testSkill
- dealbreakerSkill
- first-dollarSkill
- ground-truthSkill
- kill-or-buildSkill
- litmusSkill
- prove-itSkill
- reality-checkSkill
- receiptsSkill
- shark-boardSkill
- signalSkill
- straight-talkSkill
- truth-serumSkill
The file
We split this skill into the rules it gives an agent
8 rules. 0 are common, 3 are distinctive, and 5 are very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 25
Read ${CLAUDE_SKILL_DIR}/../receipts/rules.md first.
- Distinctive: line 35
Ask who it is for and the yes the founder wants, if not given.
- Very specific: line 40
If there is a document from them (an RFP, a grant call, an email, a job post), read it first, pull out exactly what they ask for, and save …
- Very specific: line 65
If nothing about demand is MEASURED, the pitch says so honestly through its wording: "we are looking for three firms to run a pilot", not …
- Very specific: line 76
Write PLANNED claims as goals ("we expect", "the pilot will test"), never as results.
- Distinctive: line 104
No simulated buyer, board member or model output is ever quoted, counted or described as a customer, a user, an advisor or market research …
- Very specific: line 107
No invented advisors, partners, logos, client names or numbers.
- Distinctive: line 109
No market size without a source.
“Common” is measured: kept by at least 6 in 10 of the 3 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.
Files
handshake
A plan is for the founder. A pitch is for one person and one yes. This skill writes it from what is actually known, so the founder can send it without having to defend a number they cannot back.
What the user typed after /handshake: $ARGUMENTS
Read ${CLAUDE_SKILL_DIR}/../receipts/rules.md first. The ledger tool:
python3 "${CLAUDE_SKILL_DIR}/../receipts/ledger.py" <gate | scan | summary>Below, LEDGER means exactly that.
Step 1: who, and what yes
Ask who it is for and the yes the founder wants, if not given. Name the
receiver: a customer, a funder, an investor, a lender, someone inside their
business, or the people who will use it. Read
${CLAUDE_SKILL_DIR}/receivers.md for what that receiver needs to see.
If there is a document from them (an RFP, a grant call, an email, a job post),
read it first, pull out exactly what they ask for, and save it as
founder/opportunities/<name>.md (shape:
${CLAUDE_SKILL_DIR}/../first-dollar/opportunity.md).
Push the ask down to the smallest real yes (rule 7). If the founder wants to ask a stranger for a big upfront fee, say plainly that a time-boxed pilot on the receiver's own numbers is more likely to get a yes, and offer both.
Step 2: what the folder can support
LEDGER check
LEDGER gate --audience <public | anonymised | permission>Pick the audience from how the pitch will travel: public for anything that
could be forwarded, anonymised for a private send, permission only if
every ASK FIRST claim's person said yes in founder/evidence/people.md.
Build only from what gate prints. If the receiver needs something the folder
cannot support, do not write around it: list it under Gaps at the end of
the output, with how to get it (/receipts job 4 goes deeper). Never
fill a gap with a guess, a panel quote or a board opinion.
If nothing about demand is MEASURED, the pitch says so honestly through its
wording: "we are looking for three firms to run a pilot", not "firms are
lining up".
Step 3: write it
- The one sentence, the three sentences, the three-minute outline.
- The receiver's format from
receivers.md. Answer in the receiver's own headings when they gave you a call or RFP. Respect their limits (count characters for the NSF sections and say the count). - Cite every fact with its claim id, like
[c04], in the working draft. WritePLANNEDclaims as goals ("we expect", "the pilot will test"), never as results. - A short What could go wrong section in plain words. Receivers trust it.
4b. Check the pitch stands on four legs (Ready, Fire, Aim, rule 11 in
rules.md): one big idea, the big benefit to this receiver (about them, not you), one big promise joining the two, and proof. The proof leg comes only from the claimsgateprinted; with noMEASUREDproof, the promise is the pilot's terms ("if it does not hit the line we set together, you owe nothing"). - Gaps: what this pitch would need to be stronger, and the cheapest way to get each.
Save it as founder/outputs/<YYYY-MM-DD>-<receiver>.md.
Step 4: check it before it goes
LEDGER scan founder/outputs/<file>.md --audience <the same audience>scan refuses any cited claim the audience may not see (a SIMULATED or
PRIVATE one, an ASK FIRST without permission) and flags proof words
("validated", "proven", "strong demand") when no real buyer has paid or signed.
Fix every problem it prints and run it again until it is clean. Then make a
send-ready copy without the [c..] ids if the founder wants one, and keep the
cited draft.
Rules
- No simulated buyer, board member or model output is ever quoted, counted or described as a customer, a user, an advisor or market research from real people.
- No invented advisors, partners, logos, client names or numbers. "Advisors in
accounting" needs a real advisor in
people.mdwho agreed to be listed. - No market size without a source; otherwise write it as an estimate with the method.
- Sending is the founder's job. Draft the email or message; do not send it.
Output
founder/outputs/<date>-<receiver>.md (plus a clean copy if asked), and
founder/opportunities/<name>.md if there was a call or RFP. End with the one
sentence, the gaps, and the next action: who to send it to this week, and what
to log with /receipts when they answer.
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
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.

