Resume-Builder CLAUDE.md

A project instruction file for an AI resume and cover-letter builder. It defines how applications should use job-description evidence, preserve the candidate’s facts and voice, and separate research, writing, checking, and editing.

In plain words
What is it for?
It is for tailoring resumes and cover letters to job descriptions, checking evidence for requirements, auditing writing style, and screening candidate fit.
Why use it?
It gives the agent clear rules for producing truthful, targeted applications instead of relying on one general recruiting score.

Instructions file

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/jananthan30/resume-builder/claude-md
Clone the repo
git clone --depth 1 https://github.com/jananthan30/Resume-Builder
Per session 7,624 This file is loaded in full into every session.
When invoked 7,624 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.07624 $0.07624
Opus 5 $0.03812 $0.03812
Sonnet 5 $0.01525 $0.01525
Haiku 4.5 $0.00762 $0.00762

Measured yesterday against content hash 541ea9f9a479, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

Resume-Builder CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

CLAUDE.md · 634 lines

How it starts

The opening of the file, as written. The whole thing — 634 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Resume Builder - Project Context

Overview

AI-powered resume and cover letter generator that tailors applications to specific job descriptions with native role separation and evidence-based matching:

  • Authenticity: Never invent facts; resumes must be interview-true
  • Human Voice: Brevity, burstiness, plain language — hard gate via human_voice_audit.py
  • Evidence Match: Requirement-to-evidence matching via evidence_match.py (75%+ target, zero must-haves lacking evidence). This replaced ATS/HR scoring as the authoritative measure. No universal ATS score exists across recruiting systems, so the product answers "which of this job's requirements does the resume evidence, and with what excerpt"
  • Legacy ATS/HR: ats_scorer.py / hr_scorer.py remain for continuity and the comparison table. Diagnostic only — never present them as the result
  • Candidate Fit: The configured master must score at least 50 (the default bar; candidate_fit_preflight.CANDIDATE_FIT_THRESHOLD) with zero hard knockouts against the exact JD before any resume development
  • Safety: Researcher, Writer, Auditor, and Editor have distinct least-authority contracts; finalization is ordered

Editorial priority (never invert): Authenticity → Human voice → HR impact → evidence coverage. A 75% evidence match with human prose beats a 90% keyword-stuffed draft — stuffing raises keyword counts while lowering evidence quality.

Absence is never failure. A requirement with no supporting resume span is NO_EVIDENCE_FOUND, and a hard gate with nothing behind it is UNVERIFIED. Neither means the candidate lacks the qualification, and neither may be reported as if it did. Only explicit contradictory evidence (an expired licence, say) produces FAIL.

Quick Start

Option 1: Claude Code Custom Commands (Recommended)

# Full application package (resume + cover letter) - Dual optimized
/resume [paste job description]

# Resume only - Dual optimized
/tailor-resume [paste job description]

# Cover letter only
/cover-letter [paste job description]

# Explicit native four-role workflow (authorized resume.md draft only)
/resume-team [paste job description]

# Deterministic candidate-fit gate only
/job-fit [paste job description]

Read the full file on GitHub · 634 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 634 lines · 7,624 tokens per session scan A 541ea9f9a479

Subscribe to this mod's changes

Resume-Builder CLAUDE.md is an instructions file published in the GitHub repository jananthan30/Resume-Builder (76 stars, last pushed 17d ago), licensed MIT. It adds 7,624 tokens to every session, about $0.0381 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.