send-outreach
Skill12 starsChanged 11 days agoMITView on GitHub
This skill should be used when the user wants to send a cold outreach or follow-up email to an HR contact at a company. Triggers on phrases like "send outreach to [name] at [company]", "email HR at [company]", "send cold email to [name]", "outreach [company]", "send mail to [name] [email]", "reach out to [company] HR", "email [name] from [company]", "send outreach email", "send this week's outreach emails", "send weekly outreach", "send the drafted emails".
1 other project uses a version of this.
Adopted by
Most recently taken up by
- prateekjain/ChiefOfStafftheir own version, first seen Mar 16, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- prateekjain/ChiefOfStaff★ 4same 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 repository12 other files it gives its agents.
- cover-letterSkill
- fit-checkSkill
- mark-rejectedSkill
- new-jobSkill
- open-hot-jobsSkill
- reject-jobSkill
- remove-hot-jobSkill
- resume-tailorSkill
- searchSkill
- test-runSkill
- update-hrSkill
- update-statusSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
15 rules. About 0 are common, 5 are distinctive, and 10 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
Distinctive · estimate
This author's own choices. Worth a look.
10
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 25
If the user hasn't provided all required info, ask for it before running.
- Very specific: line 31
Always send a test to your own address (recipient_email in config.py) first so the user can approve the email:
- Very specific: line 47
Add --cc "[CC_EMAIL]" if a CC was provided.
- Very specific: line 49
Add --why "..." with 1 specific, true reason this company appeals (their product, domain or tech, ideally tied to SMARTS/post-trade).
- Very specific: line 55
If not in tracker → sets role to [Company] opportunities and uses email_cold_outreach_spontaneous_template.txt
- Distinctive: line 62
Never send to real recipient without explicit user approval: after the test
- Very specific: line 63
If the script errors on missing attachment, check that resume/ folder has both PDFs
- Very specific: line 64
If template not found, check that emailoutreach/email_cold_outreach_template.txt exists
- Distinctive: line 67
Do not hardcode HR contact details anywhere in code or skill files
- Distinctive: line 76
When triggered by "send this week's outreach emails", "send weekly outreach", "send the drafted emails" — or automatically at the end of …
- Very specific: line 79
Check claude-job-agent\outreach_drafts\ for *_draft.txt files.
- Distinctive: line 92
Do NOT send, and instead note as skipped in the receipt (see Step 4), if:
- Very specific: line 99
After a draft sends successfully, delete that _draft.txt file so it doesn't appear again next week.
- Very specific: line 102
After sending (or attempting) all drafts, send a plain-text receipt to your own address (recipient_email in config.py) listing, for every …
- Very specific: line 111
If outreach_drafts/ doesn't exist or has no draft files, fall back to the standard single-email flow and ask the user for --name, --email, …
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Send HR Outreach Email
Send a personalised cold outreach email with resume + portfolio attached using send_outreach_emails.py.
Script location
claude-job-agent\send_outreach_emails.pyWhat you need from the user
| Argument | Required | Notes |
--name | Yes | HR contact full name (e.g. "Andrea Smith") |
--email | Yes | HR contact email address |
--company | Yes | Company name exactly as in tracker (used for role lookup) |
--cc | No | CC email address (e.g. second recruiter) |
If the user hasn't provided all required info, ask for it before running.
Steps
1 — Preview first (dry run to user's inbox)
Always send a test to your own address (recipient_email in config.py) first so the user can approve the email:
cd claude-job-agent
echo yes | python send_outreach_emails.py --name "[NAME]" --email "YOUR_EMAIL@gmail.com" --company "[COMPANY]"Show the preview output to the user and ask: "Looks good? Send to the real recipient?"
2 — Send to real recipient (only after user confirms)
cd claude-job-agent
echo yes | python send_outreach_emails.py --name "[NAME]" --email "[EMAIL]" --company "[COMPANY]"Add --cc "[CC_EMAIL]" if a CC was provided.
Add --why "..." with 1 specific, true reason this company appeals (their product, domain or tech, ideally tied to SMARTS/post-trade). It completes "What drew me to {company} specifically is ...". Leave it out rather than write something generic; the line is then dropped.
What the script does automatically
- Looks up the role from
List.xlsx(matches company name, status =done) - If found in tracker (role known/overridden) → status=done means already applied, so it uses
email_followup_template.txt("I recently applied for the {role} role...") — never the agency-flavored cold outreach template. If the role title mentions AI/IA/ML/LLM/GenAI/NLP/agentic (seeAI_ROLE_PATTERN), it usesemail_followup_ai_template.txtinstead, which leads with the LLM projects - If not in tracker → sets role to
[Company] opportunitiesand usesemail_cold_outreach_spontaneous_template.txt - Attaches both PDFs from
resume/folder - Shows full preview before sending
Rules
- Never skip the test send step — always preview to user's inbox first
- Never send to real recipient without explicit user approval after the test
- If the script errors on missing attachment, check that
resume/folder has both PDFs - If template not found, check that
emailoutreach/email_cold_outreach_template.txtexists - LinkedIn InMail variants also live in
emailoutreach/:linkedinoutreach_followup_template.txt(checking status on an application) andlinkedinoutreach_coldintro_template.txt(plain cold outreach, no application) - Engineer / EM outreach, close domain match, unclear who hires (non-AI role; fraud, AML, risk, compliance, surveillance): don't use the script's template. Fill
emailoutreach/email_cold_outreach_domain_match_engineer_template.txtby hand (its HOW TO FILL block lists the slots), test-send it to self viasend_email()insend_outreach_emails.py, then send on approval - Do not hardcode HR contact details anywhere in code or skill files
- If the job listing URL returns 404 or any error (expired/taken down), do not send the outreach email — the role no longer exists. Either switch to the spontaneous template or skip entirely and notify the user.
Weekly draft mode — "send this week's outreach emails"
Per user instruction (2026-08-17): drafts to known/verified HR contacts are sent automatically, no per-email confirmation. This applies to drafts produced by the weekly-hr-search routine, and to any future "draft outreach then send" workflow unless the user says otherwise.
When triggered by "send this week's outreach emails", "send weekly outreach", "send the drafted emails" — or automatically at the end of the weekly-hr-search routine right after drafts are written:
Step 1 — Read saved drafts
Check claude-job-agent\outreach_drafts\ for *_draft.txt files.
Parse each file:
TO: {email}
SUBJECT: {subject}
---
{body}Step 2 — Send automatically
For each draft, send via send_outreach_emails.py (or an equivalent direct SMTP send of the draft's own SUBJECT/body, which was built from email_followup_template.txt (or email_followup_ai_template.txt for AI roles) for tracked roles or email_cold_outreach_spontaneous_template.txt otherwise; never email_cold_outreach_template.txt) — no confirmation prompt, no dry-run-to-self step.
Do NOT send, and instead note as skipped in the receipt (see Step 4), if:
- Finding a contact or email address: use Hunter, then RocketReach (same order as the update-hr skill), before any guessed pattern. Apollo is not used (Free plan blocks it).
- The contact's email is flagged undeliverable/invalid by Hunter (or any verifier). Confirmed 2026-08-17:
cecile.grondin@netatmo.comwas flagged "undeliverable" by Hunter, sent anyway, and did in fact bounce/get flagged by Netatmo's mail server — so trust the verifier's "invalid/undeliverable" status and skip rather than guess-and-send. - A guessed email has no verification at all (Hunter wasn't able to check it, e.g. accept-all domain) — still send it, but flag it clearly as unverified in the receipt so the user knows it's a lower-confidence guess. This is different from a confirmed-bad flag above.
- The job listing URL returns 404 or any "no longer available" signal — skip, do not send, note it in the receipt.
Step 3 — Clean up sent drafts
After a draft sends successfully, delete that _draft.txt file so it doesn't appear again next week.
Step 4 — Send a receipt email (mandatory, every run)
After sending (or attempting) all drafts, send a plain-text receipt to your own address (recipient_email in config.py) listing, for every email actually sent:
- Company name
- Role
- HR contact name + email sent to
- Any caveat (e.g. "unverified email, watch for bounce")
And for anything skipped: company, reason skipped.
Fallback — no drafts folder or empty
If outreach_drafts/ doesn't exist or has no draft files, fall back to the standard single-email flow and ask the user for --name, --email, --company.
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.

