Cavalry

AI-powered job search: fetch from 22+ job boards, classify with your LLM, autofill ATS forms. BYOK — fully local. No tracking.

Cavalry simultaneously searches major remote job boards, open aggregators, and RSS feeds, scores every posting against your resume and preferences using an LLM, and autofills standard applicant tracking forms. All data stays in your browser -- no Cavalry servers, no analytics, no tracking. HOW IT WORKS 1. Set your profile (resume, location, preferences) and LLM API key. 2. Click "Fetch and Classify" -- Cavalry queries your enabled job sources, pulls listings, and sends them in small concurrent batches to your LLM for scoring. 3. Jobs appear in your queue sorted by composite score, with tier badges (A/B/C/Discard). 4. Apply with one click, then use the autofill engine to complete the application form. THREE-AXIS SCORING Every job receives three 0-100 scores, blended into a composite: WANT -- How well the role matches your signal weights (salary, remote flexibility, company type) with dealbreaker flags. QUALIFIED -- How well your resume aligns with the job description. LIKELY -- Market competitiveness: seniority requirements, clearance needs, and other barriers. Jobs are bucketed into A/B/C/Discard tiers based on configurable thresholds. TUNABLE SIGNAL WEIGHTS Add, remove, or rename as many signals as you want. Each gets a 1-10 rating and an optional dealbreaker flag that triggers immediate discard. The 10 default signals are a starting point -- delete them all or build your own from scratch. Defaults include: Enterprise Gear, Skill Building, Low Travel, BS Job, Morally Questionable, Non-Profit, Religious Organization, Active Security Clearance, Heavy Seniority, Short-Term Contract. NATIVE JOB SOURCES Cavalry handles job discovery across a broad ecosystem by categorizing access models to protect your privacy and pipeline speed: Open Aggregators and Feeds: Access popular tech news feeds, open remote job boards, and startup portals immediately without needing external credentials. Bring-Your-Own-Key (BYOK) Platforms: Connect professional networks, federal hiring boards, and localized career portals directly using your personal developer API keys. Cloud Extractor Integrations: Tap into advanced web-scraping runtimes for major venture capital lists and Web3 job boards using a personal cloud token. Enable only the sources you want, and set per-source result limits. ATS AUTOFILL The content script activates automatically on nine major Applicant Tracking Systems (including Workday, Greenhouse, Lever, Ashby, and iCIMS). It scans form fields, sends field labels to your LLM for matching against your personal Answer Bank, and fills in the values locally. ANSWER BANK Store reusable Q and A pairs in structured categories (Work History, Education, Skills, Certifications, Standard Answers). Populate via manual entry, LLM generation from your resume, or CSV import. Lock finely-tuned answers to prevent LLM overwrites. ATS AGGREGATOR Discover organizations based on the software they use to manage applicants, and poll their career pages for new postings using background Chrome alarms. Includes an employer discovery tool that generates ranked company suggestions based on your resume. DATA AND PRIVACY Fully local: All data (profile, resume, jobs, settings, API keys) lives in IndexedDB. No Cavalry servers exist. No tracking: No analytics, no telemetry, no third-party SDKs, no crash reporting. You control the LLM: Default is DeepSeek. Swap to OpenAI, Anthropic, Ollama, LM Studio, or any OpenAI-compatible endpoint. You control the sources: Toggle each independently. API keys stored locally, only sent to services you configure. One-click backup: Export your entire config as JSON. Factory reset wipes everything clean. GETTING STARTED 1. Settings tab: enter your LLM API key (or point to your own local endpoint) and click Test Connection. 2. Profile tab: fill in name, email, resume text, location, and core preferences. 3. Prefs tab: toggle your preferred free sources, or add API keys for developer platforms. 4. Settings tab: rate signal weights and flag your personal dealbreakers. 5. Queue tab: click Fetch and Classify. REQUIREMENTS Your own LLM API key is required (DeepSeek, OpenAI, Anthropic, or a local model). Cavalry does not provide one. Some job sources require individual developer API keys (free tiers are widely available for public platforms). Scraping-backed sources require an independent cloud extraction platform token. Uses standard host permissions to query external APIs and inject the local autofill layer on target career sites. SUPPORT Email: cavalrysupport@benw.net Support the developer: https://buymeacoffee.com/benthesoundguy