Market Aligner

Market Aligner is one multi-profile product. It collects and normalises vacancy data, applies deterministic viability and eligibility rules, performs bounded semantic assessment, and ranks opportunities against an evidence ledger selected by profile_id.

Profile data, credentials, generated applications, collected data, caches, and runtime receipts live outside this repository under MARKET_ALIGNER_DATA_HOME.

The application-automation component is logically inside the product boundary but is separately certifiable. This repository currently contains only its versioned interface; the active implementation will be imported after its protected upstream freezes and is sealed.

Development status

This is the canonical reconciliation lineage. Audited predecessor trees remain read-only until every useful component has been adopted, tested, and recorded in the migration ledger.

Implemented local slices include:

• 34 registered board adapters, an uncapped parallel collector, durable restart state and the complete Scrapling sidecar protocol;
• generic opaque profiles, an external evidence ledger and loss-conscious legacy importers;
• deterministic normalisation, deduplication, viability, eligibility and scoring;
• validated semantic extraction/evidence-alignment schemas and content-bound LLM receipts;
• an opportunity-before-research database gate, leased research workers and cited dossiers;
• profile-scoped ranking, skill-frequency and interactive opportunity/fit reports;
• a local service layer and provisional, versioned JAA contracts.

Fit is always reported as uncalibrated. This is a ranking heuristic, not a hiring probability.

Local commands

export MARKET_ALIGNER_DATA_HOME="$HOME/.local/share/market-aligner"

market-aligner profiles list
market-aligner profiles create-synthetic
market-aligner profiles import --format evidence-led --source /private/profile.yaml
market-aligner assess --profile-id prf_<opaque-id> --request /private/request.json

Live collection configuration is external and injected into adapters. The automatic Scrapling fallback uses static then dynamic fetching; stealth/challenge-solving capabilities require an explicit source policy. Final submission and legal consent are never authorized by Market Aligner.