The architecture
32 specialized agents.
One career assistant.
Most AI job tools ship one chatbot in a trench coat. NovaHunt is a tier hierarchy of small, focused agents — each one auditable, each one with a hard spend cap, each one doing exactly one job.
The constellation
All 32 agents, at a glance.
VANGUARD at the center. Eight directors radiate out. Twenty-one specialists cluster near their director. Three safety agents (dotted lines) govern every call.
Sourcing
HARVESTER · POACHER · DRAGNET · SIEVE · CRIER
Scoring
PARAGON · DOWSER · APPRAISER · JUDGE · CALIBRATOR
Learning
SCOUT
Documents
SCRIVENER · REDACTOR · OVERTURE · MUSE · PERSONA · STEWARD
Applying
EMISSARY · PORTAL · DISPATCHER · WITNESS
Follow-up
ENVOY · CADENCE · LISTENER
Strategy
STRATAGEM · CARTOGRAPHER · REFLECTOR · AUDITOR · EXPLORER
Safety layer
WARDEN · TURNSTILE · ABACUS
Orchestrator
VANGUARD
VANGUARD at the center orchestrates. Eight domain directors radiate out in the inner ring. Twenty-one specialists cluster near their director. Three safety agents (dotted lines) govern every call. Read the architecture →
The design principle
One monolithic LLM that does everything is impossible to audit, impossible to bound, and impossible to improve. NovaHunt splits the job into 32 focused agents organized in three tiers. Each one is independently versioned, independently rate-limited, and independently swappable. If APPRAISER's scoring drifts, we replace just APPRAISER — not the whole assistant.
Underneath them sits a deterministic safety layer (WARDEN, TURNSTILE, METER, CANARY) that is written in pure Python and contains zero LLM imports. Agents propose, the safety layer disposes. That's what makes the spend caps real.
Tier 1
The conductor
One agent that orchestrates every other agent and assembles your daily brief.
VANGUARD
ConductorEvery agent's queued briefing items from the last 24h
Your daily brief: hot matches, follow-ups owed, replies that need you
$0.05/run
Tier 2 — 8 directors
Domain directors
Each one owns a stage of the search pipeline and delegates to its specialists.
HARVESTER
Sourcing DirectorYour target roles, locations, and enabled job boards
Newly fetched + deduped postings across every connected source
$0 (no LLM — pure HTTP)
PARAGON
Matching DirectorNewly sourced postings and your profile
Scored, ranked, and rationale-attached matches
$0.50/batch (max 12 jobs)
SCRIVENER
Documents DirectorA specific job description and your master profile + voice
Tailored resume + cover letter as new document versions
$0.10/job
EMISSARY
Application DirectorA drafted application and your channel preference
Queued ATS submission or composed recruiter email
$0.04/application
ENVOY
Follow-up DirectorEvery application's stage and time-since-submit
Overdue-follow-up flags for VANGUARD to escalate
$0 (no LLM)
STEWARD
Assets DirectorYour resume + cover letter library and last-edit timestamps
Staleness flags, version recommendations, missing-template warnings
$0 (no LLM)
STRATAGEM
Strategy DirectorYour full search history week-over-week
Pattern-spotting briefings ("your offers come from sub-50 person companies")
$0.05/week
CARTOGRAPHER
Coverage DirectorWhich sources produce your highest-scoring matches
Source-mix recommendations and gap-coverage suggestions
$0 (no LLM)
Tier 3 — 21 specialists
Focused specialists
Each does exactly one thing — score a job, write a cover letter, classify a reply — and does it cheaply.
POACHER
ATS specialistPublic ATS endpoints (Greenhouse, Lever, Ashby, Workable)
Raw postings normalized to one schema
DRAGNET
Aggregator specialistAdzuna + JSearch (LinkedIn/Indeed/Glassdoor proxy)
Raw postings normalized to one schema
SIEVE
Dedup specialistRaw postings from every source this run
Deduped jobs keyed by (source, source_id) + URL hash
CRIER
NotifierHot matches and digest payloads
Delivered email via Resend MCP
DOWSER
Prefilter specialistCandidate jobs vs your profile vector
Top-K shortlist when the candidate pool > 50
APPRAISER
Per-job scorerOne job + your profile
Normalized score [0,1] + 3-5 weighted factors
$0.02/job
JUDGE
Rationale specialistAn APPRAISER score above 0.60
Plain-English rationale + gaps + top-applicant tips
$0.04/job
SCOUT
Local learnerYour saved + dismissed labels (sklearn TF-IDF + LogReg)
Personalized fit score that re-ranks the list — $0 marginal cost
CALIBRATOR
Quality auditorScored matches vs your real saves/dismisses
Score-calibration drift signals
REDACTOR
Resume tailorerYour master resume + this specific JD
Tailored resume emphasizing JD keywords
$0.04/job
OVERTURE
Cover-letter writerJD + your achievements + voice notes
Cover letter in your tone
$0.04/job
MUSE
Voice analystYour uploaded resumes + cover letters
Voice fingerprint so future drafts sound like you
PERSONA
Top-applicant briefThis JD + your profile
Archetype + ✓/○ checklist + differentiator
$0.03/job (cached)
PORTAL
ATS submission stagingDrafted application + ATS metadata
Hands you a deep-link to submit (full automation on roadmap)
DISPATCHER
Email submission specialistDrafted application + recruiter address
Sent email via Resend with your reply-to
WITNESS
Confirmation loggerConfirmation reference (ATS ref / message-id)
Verified application stage flip to 'applied'
CADENCE
Follow-up drafterSubmitted application + days elapsed
Tone-appropriate follow-up email draft
$0.02/draft
LISTENER
Reply classifierA recruiter's inbound reply
Classification (interview / rejection / etc.) + extracted action
$0.02/reply
REFLECTOR
Outcome analystApplications and their final stages
Conversion-rate analysis per role / per source
$0.05/week
AUDITOR
Process auditorEvery agent_run for the week
Spend + latency + failure breakdown
EXPLORER
New-source scoutSource performance + jobs you missed elsewhere
Recommendations for new boards to enable
ABACUS
Spend anomaly detectorHourly + daily spend rollups across every user
Anomaly flags that halt spending until you approve
The safety layer
Deterministic guardrails
Zero LLM imports. The safety layer is what makes the spend caps and consent tiers actually enforceable — an agent cannot reason its way past a deterministic Python validator.
WARDEN
Deterministic safety validatorZero LLM imports. Pure Python validators that gate every agent action against the Constitution. If WARDEN says no, the action never happens — the agent can't talk it into yes.
TURNSTILE
Pre-spend authorization gateEvery Claude call asks TURNSTILE first. It checks your daily and monthly caps, your plan tier, your active spend holds, and your consent tier for that action class. Denial returns a 402 'approval needed' — the agent halts cleanly.
METER
Token + cost ledgerRecords every Anthropic call (model, input tokens, output tokens, USD) into the agent_runs table. Drives your spend dashboard and ABACUS's anomaly detection.
CANARY
Event-rule monitorWatches the audit stream for suspicious patterns (e.g. an agent that suddenly spends 10× its usual rate) and emits halt-and-approve events to the user.
WATCHTOWER
Per-user resource governorTracks rate limits per action class (intake parses/hour, refreshes/hour, auto-applies/day) and routes denial through TURNSTILE so the user sees the same '402 approval needed' UX everywhere.
Consent tiers
GREEN / YELLOW / RED — you decide what's automatic
Every action class has a consent tier. GREEN actions run silently (e.g. extract profile from a resume you just uploaded). YELLOW actions queue for your one-tap approval (e.g. submit an application). RED actions require step-up authentication (e.g. rotate your API credentials). You can re-tier any action at any time in Settings → Consent; RED actions can never be auto-promoted to GREEN.
See it run on your own resume
Sign in, upload your resume, and the agents go to work. First refresh surfaces real matches; first save trains your personalized re-ranker. No card required to try.