Cold outreach for work
A stack of skills for your AI agent, made for people starting from zero. 103 skills and 149 project briefs — load one into Codex, Claude Code, Copilot, Gemini, or Cursor and do the thing.
npx -y skills add protosphinx/sphinxstack --skill cold-outreach-for-workAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 18 days oldThe repository was created 18 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Reach out to businesses that never posted a job. Short, specific messages in the person's voice, sent by them. Use when they say "should I just email them", "there's no job posted but I want to work there", or want to reach someone they look up to.
SKILL.md
3.1 KB, 663 tokens by cl100k_base, as published. Nobody here has run it
cold-outreach-for-work
Help someone contact strangers about work — the shop with no posting, the studio they admire, the person whose career they want. The bar is a handful of messages that are short, specific, and true, sent by them before the sitting ends.
Ground rules
- They send everything from their own accounts, under their own name. You draft with them and never message anyone as them.
- Every claim in a message is true and every bit of specificity is real — if the message says they admire the recipient's work, they must be able to name what and why. No flattery templates.
- Respect the recipient's time: under 120 words, one clear small ask, no attachments unless asked, no guilt-tripping, and a hard cap of one follow-up per person ever. Public contact info only — the address on the website, the shop's counter, a public profile. No scraping, no personal numbers passed around.
Targets and the ask
List five to ten real targets with them, each with a reason they can say out loud — walks past it daily, uses the product, admires the work. For each, pick the one ask, matched to what a stranger can grant: small businesses get "do you ever take on part-time help / could I come in and introduce myself"; admired people get "could I ask you two questions about how you started" — advice first, since asking a stranger for a job usually gets silence but asking for fifteen minutes often gets a reply. Find the right channel per target: the owner's email on the site, the contact form, or simply walking in (that path is apply-in-person).
Draft and workshop
Draft each message in their voice, four parts in under 120 words: who they are in one line (student status is enough; do not tell strangers a minor's exact age), the specific true reason for writing to this recipient, the one ask, and an easy out ("no worries if not"). Subject line plain and honest: "question from a student" beats anything clever. Then workshop: read it aloud, cut every line that could be sent to anyone else, cut every word they would not say to the person's face. The message ships when it is specific enough that only this recipient could have received it.
Send and track
They review and send from their account while you are together. Log each message in their tracker (job-application-tracker builds one) with a single follow-up date about a week out; follow-up-after-applying handles that nudge, and silence after it means done, no exceptions. When a reply comes: answer fast, say thanks, and take whatever was offered seriously — a fifteen-minute chat with someone doing the work is the actual prize, and interview-prep can rehearse it.
Done
- Five to ten targets, each with a true reason and one small ask
- Messages under 120 words, in their voice, sent by them
- Every send logged with one follow-up date and a reply plan
What ships with it: 1 file
232 B alongside SKILL.md
agents/
- openai.yaml232 B
Gives 0 of the 12 instructions most sales audience skills give in 663 tokens
Counted across 401 of the 401 authors here whose files we hold, read 2026-08-07
- Read product marketing context before asking questionsin 21 of 401, across 11 files
- Acknowledge competitor strengths honestlyin 18 of 401, across 7 files
- Start every page with a summaryin 15 of 401, across 4 files
- Use a single, low-friction call to actionin 15 of 401, across 7 files
- Create a single source of truth for each competitorin 14 of 401, across 3 files
- Make each follow-up email add new valuein 11 of 401, across 5 files
- Cut any sentence that does not drive a replyin 10 of 401, across 4 files
- Tie personalization directly to the problemin 10 of 401, across 4 files
- Write paragraph comparisons for each dimensionin 9 of 401, across 3 files
- Link between related competitor pagesin 9 of 401, across 3 files
- Keep subject lines short and lowercasein 9 of 401, across 3 files
- Define ideal customer profile from top customersin 9 of 401, across 3 files
Said here and by no other author read
- Let them send all messages from their own accounts
- Make every specific claim in a message verifiably true
- List five to ten real targets with true reasons
- Draft each message in their voice
- Cut generic lines anyone could send
- Have them review and send messages while together
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.