cover-letter-writer
Skill4 starsChanged 7 months agoMITView on GitHub
Write and iteratively refine cover letters using a 3-agent debate team (2 Opus critics + 1 Sonnet judge). Use this skill whenever a job application allows cover letter upload, whenever the user asks to write or improve a cover letter, or whenever the job-screener skill returns APPLY for a position. The must-write rule (if enabled in config) applies: if the application system has a cover letter upload field, a cover letter MUST be written. Trigger on: "write cover letter", "cover letter", any application form with a cover letter upload option, or after a job-screener APPLY decision.
We split this skill into the rules it gives an agent
11 rules. 0 are common, 10 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 9 edited versions we read keep these.
10
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 9
Every cover letter goes through a rigorous 3-agent debate review process before submission.
- Distinctive: line 13
If this is the first job-application skill invoked in this session, run the self-check skill first.
- Very specific: line 17
If config.cover_letter.must_write is true: whenever the application system allows uploading a cover letter, you MUST write one.
- Distinctive: line 21
Read from three sources:
- Distinctive: line 33
Mirror JD language exactly: If the JD says "Cursor and Codex," use those exact words in paragraph one.
- Distinctive: line 35
Specific project mechanics, not generic claims: Never write "I am detail-oriented." Instead: "issuer ID and plan ID as only join key, …
- Distinctive: line 37
Root cause → fix → lasting improvement arc: Use the user's best debugging/obstacle story from their projects.
- Distinctive: line 80
After writing the initial draft, spawn three agents in parallel:
- Distinctive: line 141
Run all 3 agents in parallel for speed
- Distinctive: line 142
After receiving feedback, revise the draft addressing the top issues
- Distinctive: line 149
After the cover letter is finalized, generate a PDF using reportlab:
“Common” is measured: kept by at least 6 in 10 of the 9 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
Cover Letter Writer
Every cover letter goes through a rigorous 3-agent debate review process before submission. This isn't about producing generic letters — it's about crafting dense, specific, JD-mirrored arguments for why the applicant is the right fit.
Pre-Flight
If this is the first job-application skill invoked in this session, run the self-check skill first. The cover letter writer needs the user's latest project details, resume, and background — stale data produces generic letters.
Must-Write Rule
If config.cover_letter.must_write is true: whenever the application system allows uploading a cover letter, you MUST write one. No exceptions. Check the application form for a cover letter upload field before skipping this step.
User Background (for content)
Read from three sources:
config.yml—config.projectsfor structured project data,config.primary_strength+config.primary_strength_evidencefor the user's key differentiator- Auto-memory (
user_background.md) — for detailed project mechanics, narrative arcs, and debugging stories config.registered_files— check forproject_doc,portfolio,writing, orcertificatetype files. Read them for additional material that could strengthen the cover letter. For example, a portfolio PDF might have project details not captured in config.
The config provides the facts; the memory provides the storytelling material. Use both.
Writing Standards
What makes a great cover letter:
- Mirror JD language exactly. If the JD says "Cursor and Codex," use those exact words in paragraph one. If they say "onboarding/offboarding," use that phrase. The reader should feel like the letter was written specifically for their posting — because it was.
- Specific project mechanics, not generic claims. Never write "I am detail-oriented." Instead: "issuer ID and plan ID as only join key, silent record drops." The specificity IS the evidence.
- Root cause → fix → lasting improvement arc. Use the user's best debugging/obstacle story from their projects. Adapt it to the JD context. This shows persistence, technical depth, and engineering discipline.
- Confident day-one action closing. End with what the applicant will DO on day one, not what they hope to learn. Use "I will" or "I am ready to," never "I would be glad to" or "I hope to."
- Under
config.cover_letter.max_wordswords (default 330). Density beats length. Every sentence must earn its place.
What to avoid:
- Generic personality claims ("I am a hard worker," "detail-oriented," "team player")
- Advanced ML techniques for ops-focused roles — sounds overqualified and disconnected
- Passive language ("glad to own") — use "ready to own" or "I will"
- Stating obvious location info (if user is at a university in the job's city)
- Listing tools as skills without a concrete use case in the body
- Overly long sentences or paragraphs — keep it punchy
Also incorporate any user-specific writing preferences from feedback_cover_letter.md in auto-memory.
Cover Letter Structure
[Header: Name, Email, Phone, Date]
(Use config.personal.* for header fields)
Dear Hiring Manager,
[P1: Hook — connect the user's most relevant experience directly to the JD's
primary need. Mirror JD language. Establish credibility in 2-3 sentences.]
[P2: Debugging/obstacle story — adapt the root cause → fix → improvement arc
to the JD's domain. Show persistence and engineering discipline.]
[P3: Supporting projects — 1-2 more projects that reinforce the match.
Connect to specific JD requirements. Technical specifics, not summaries.]
[P4: Motivation + day-one action — why this specific role/company excites
the user, and what they will do immediately upon starting.]
Sincerely,
[config.personal.first_name] ([config.personal.preferred_name]) [config.personal.last_name]3-Agent Debate Review Process
After writing the initial draft, spawn three agents in parallel:
Agent 1: Advocate (Opus)
You are the Advocate reviewing a cover letter for a job application.
Your job is to identify and defend the letter's STRENGTHS:
- Where does it effectively mirror JD language?
- Which specific details are most compelling?
- How well does the narrative arc work?
- Is the closing strong and actionable?
Score the letter 1-10 and explain what's working. Be generous but honest —
if something is mediocre, don't call it great.
JD: [paste JD]
Cover Letter: [paste draft]
User's Background: [paste from user_background.md and config.projects]Agent 2: Devil's Advocate (Opus)
You are the Devil's Advocate reviewing a cover letter for a job application.
Your job is to find WEAKNESSES and GAPS:
- What JD requirements are NOT addressed?
- Where is the language generic instead of specific?
- Are there any claims without evidence?
- Does anything sound overqualified or disconnected from the role?
- Is the word count under 330? Is every sentence earning its place?
- Would a hiring manager for THIS specific role be convinced?
Score the letter 1-10 and list the top 3 most important improvements.
Be tough but constructive.
JD: [paste JD]
Cover Letter: [paste draft]
User's Background: [paste from user_background.md and config.projects]Agent 3: Judge (Sonnet)
You are the Judge synthesizing two peer reviews of a cover letter.
You will receive an Advocate's defense and a Devil's Advocate's critique.
Your job is to:
1. Weigh both perspectives fairly
2. Give a final score (1-10)
3. Identify the single most important improvement that would raise the score
4. Decide: is this letter ready to submit, or does it need another round?
Threshold: both Opus critics should score 8+ before submission.
JD: [paste JD]
Cover Letter: [paste draft]
Advocate Review: [paste]
Devil's Advocate Review: [paste]Iteration Rules
- Run all 3 agents in parallel for speed
- After receiving feedback, revise the draft addressing the top issues
- Re-run the debate until both Opus agents score >=
config.cover_letter.score_threshold(default 8) - Max
config.cover_letter.debate_rounds_maxiterations (default 3) — if scores plateau, present to user for final decision - Track version numbers (v1, v2, v3...) so user can see the progression
PDF Generation
After the cover letter is finalized, generate a PDF using reportlab:
# Use config.cover_letter.font (default Times-Roman), letter size, 1" margins
# Header: Name centered, email | phone | date below (from config.personal.*)
# Body: justified text, proper paragraph spacing
# Save to: [config.files.cover_letter_dir]/[Company]_CoverLetter_[LastName].pdfThe PDF filename pattern: [Company]_CoverLetter_[config.personal.last_name].pdf
Other versions9 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 9
- squerne/open-career-skills★ 26same name
- jaskirat1616/grok-skills★ 22same name
- JayRHa/agent-skills★ 12same name
- JayRHa/AgentSkills★ 12same name
- art2url/career-agent-skills★ 9same name
- kyoungbinkim/give-me-job★ 6same name
- DexterslabAi/resume-genie★ 0same name
- fred-brown-3/job-hunter-ai★ 0same name
- SkillMedev/job-search-kit★ 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.
Used alongsideProjects with it also use 0-template, github-actions-builder and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
JayRHaSkill0-template
Template for creating a new Agent Skill. Copy this folder, rename it to your skill's kebab-case name, and fill in every section. Use this as the starting point whenever you author a new skill for the AgentSkills library.
Also in 3 projects that use this
JayRHaSkillgithub-actions-builder
Builds reliable, fast, and secure GitHub Actions CI/CD workflows with dependency caching, build matrices, scoped secrets, concurrency control, and reusable/composite workflows. Use this skill when the user asks to "create a GitHub Actions workflow", "set up CI", "add CI/CD", "fix a flaky/slow GitHub Actions pipeline", "add a build matrix", "cache dependencies in CI", "write a reusable workflow", "configure deployment with environments and secrets", or mentions files under .github/workflows.
Also in 3 projects that use this
xmqywxSkillmeeting-summarizer
Transform meeting transcripts into structured summaries with action items, decisions, follow-ups, and efficiency analysis. Works with any meeting format.
Also in 3 projects that use this
MehdibargachSkillokr-writer
Turn business goals into structured OKRs with measurable Key Results, linked initiatives, alignment check, and anti-goals. Takes a strategic priority and outputs ready-to-use OKRs.
Also in 3 projects that use this
pbc-osSkillsecrets-manager
Set up and configure cloud secrets management for your AI agent. Guides users through platform selection (GCP, AWS, Azure, 1Password, Doppler, HashiCorp Vault), account setup, CLI configuration, and secure credential storage.
Also in 3 projects that use this
- +7 more from JayRHa
In this repository5 other files it gives its agents.
- application-fillerSkill
- application-trackerSkill
- job-screenerSkill
- job-searchSkill
- self-checkSkill
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.

