Truthful, agent-driven job applications: one set of facts woven into every resume, cover letter, and form…
FAQ (answers come from our records)
How do I install it?
We did not find an install command for this repository (our install coverage is only about 4%). Check its README or releases page.
Who is it for?
We filed it under these tags: 开源项目. The tags come from the entry itself, not a taxonomy we imposed.
Who maintains it?
repository DevGuyRash/factloom;first catalogued 2026-10-12;last seen 2026-10-16。There is no official certification or quality rating here — we record sources and dates only.
factloom
以下摘录自该仓库的 README(片段) · 在 GitHub 读完整 README →
Truthful, agent-driven job applications. You write every fact about your work once; factloom weaves those facts into tailored resumes, cover letters, and form answers, and an AI agent (Claude Code, Codex, or any agent that reads AGENTS.md) applies to jobs from your browser while recording everything it sends.
New resume batches default to research-mode onboarding. The agent explains synthetic-content capabilities before drafting and offers adjustable research/enrichment controls or evidence-only for a real career resume. Existing batches retain their saved settings. Controls inherit across variants and tailored outputs; the CLI is unchanged. Research records preserve provenance while agreed blinded artifacts can look ordinary. Genuine source history remains intact, and external tests require company agreement and submission authorization.
| classic-blue | modern | executive |
|:---:|:---:|:---:|
| | | |
| **compact** | **sidebar** | **plain** |
| | | |
*Every resume above is built from the same facts file of a fictional person (examples/demo); only the theme differs. Themes are short YAML files you can copy and change: colors, fonts, type, corners, heading styles, and layout.*
What it does
- **One source of truth.** Each claim lives once in facts.yaml. Resume versions (variants) pick and order facts for a kind of role; themes decide the look. Fix a typo once and every resume picks it up.
- **Starts from what you have.** Put available resumes in drop/: any format, any number of versions, even a scan, photo, or LinkedIn export. ./resumes import archives and extracts them; the agent reads what no converter can and reconciles the versions with your existing evidence. It asks about material unresolved differences rather than rechecking facts you already settled. The resume-builder researches the target's actual work, recovers missing proof, and reviews role-specific drafts before delivery, with a fresh reader when available.
- **Claims stay reviewable.** Unconfirmed bullet phrases carry confirm flags that the build uses to mark the resume for review; tailoring refuses them by default. Other unresolved claims and pending approval belong in the guide's manual review items. These checks support source-based review; a generated ready status is not proof of truth or approval to send.
- **Applies for you.** Open job postings or job sites in the browser your agent controls and say "apply to the jobs open in my browser." The agent captures each posting, learns what the employer does, picks and tailors a resume, writes a cover letter when wanted, fills the form from your answers, submits or holds it, and records what it sent. It keeps going until you stop it, for weeks if you like: it rotates through searches for each target title on each job site, questions it cannot settle wait in one batch for you, and when the leads run out it widens the search and keeps watching for new postings. Where the agent's host can schedule it, it sets up the schedule itself after onboarding, unless you ask for a single session or another limit, running the repository's own prompt (./resumes template automation-prompt), a template you can change in one place.
- **Knows what to apply for.** Each resume version's guide is written from current research: the titles employers use, the seniority the facts support, pay, and saved searches that find those roles. ./resumes status flags guides due for fresh research (every 90 days by default). Each posting gets a fit score: postings where you meet fewer than half the requirements, or that score below the minimum you choose in onboarding, are skipped.
- **Keeps your clock, if you allow it.** It answers start-date and availability questions from your notice period and interview hours, counted from the day of each application; offers and books interview times inside your windows, never sooner than your notice, with the time zone written out and your calendar checked for conflicts; reminds you before each interview; thanks the interviewer afterwards; and shows interviews and deadlines first in ./resumes status.