[feat] 13-Language Heuristic Classifier & Safety Tier Engine #46

New issue

Closed

opened 2026-08-30 17:58:52 +03:00 by jalmari · 1 comment

jalmari commented

2026-08-30 17:58:52 +03:00

Owner

Parent

#44: [spec] Port MailSweep to Synchronous Rust

What to build

Implement the complete 13-European-language heuristic rules engine compiled statically with std::sync::LazyLock and regex. Enforce the inviolable Safety Tier hierarchy (Tier 0 through Tier 5) over Clutter Tiers with 100% parity against the Python classifier oracle.

Acceptance criteria

13 European language dictionaries (Nordic, Baltic, Western, Southern) and keyword regex sets compiled with std::sync::LazyLock.
Inviolable Safety Tier Precedence Hierarchy enforced: Tier 0 (Domain Immunity), Tier 1 (Human Conversations), Tier 2 (Attachments), Tier 3 (Invoices/Receipts), Tier 4 (Critical Records), Tier 5 (Shipping/Calendar).
Clutter Tiers accurately classify newsletters, surveys, social pings, and promotions.
Gmail X-GM-LABELS enriched and integrated into classification scoring.
Classifier unit test matrix passes with 100% parity against Python reference results.
cargo test, cargo fmt --check, and cargo clippy --all-targets --all-features -- -D warnings pass with 0 warnings.

Blocked by

#45: [feat] Core Domain Types, Configuration & RFC Codecs

## Parent #44: [spec] Port MailSweep to Synchronous Rust ## What to build Implement the complete 13-European-language heuristic rules engine compiled statically with `std::sync::LazyLock` and `regex`. Enforce the inviolable Safety Tier hierarchy (Tier 0 through Tier 5) over Clutter Tiers with 100% parity against the Python classifier oracle. ## Acceptance criteria - [ ] 13 European language dictionaries (Nordic, Baltic, Western, Southern) and keyword regex sets compiled with `std::sync::LazyLock`. - [ ] Inviolable Safety Tier Precedence Hierarchy enforced: Tier 0 (Domain Immunity), Tier 1 (Human Conversations), Tier 2 (Attachments), Tier 3 (Invoices/Receipts), Tier 4 (Critical Records), Tier 5 (Shipping/Calendar). - [ ] Clutter Tiers accurately classify newsletters, surveys, social pings, and promotions. - [ ] Gmail `X-GM-LABELS` enriched and integrated into classification scoring. - [ ] Classifier unit test matrix passes with 100% parity against Python reference results. - [ ] `cargo test`, `cargo fmt --check`, and `cargo clippy --all-targets --all-features -- -D warnings` pass with 0 warnings. ## Blocked by - #45: [feat] Core Domain Types, Configuration & RFC Codecs

jalmari added the

ready-for-agent

feature

labels

2026-08-30 17:58:52 +03:00

jalmari added a new dependency

2026-08-30 17:58:52 +03:00

#45 [feat] Core Domain Types, Configuration & RFC Codecs

jalmari referenced this issue

2026-08-30 17:58:52 +03:00

[feat] Mailbox Pipeline & Safety Auditor (End-to-End Orchestrator) #49

jalmari added a new dependency

2026-08-30 17:58:52 +03:00

#49 [feat] Mailbox Pipeline & Safety Auditor (End-to-End Orchestrator)

jalmari closed this issue

2026-08-30 18:15:29 +03:00

jalmari commented

2026-08-30 18:15:38 +03:00

Author

Owner

Implemented 13-language heuristic classifier and safety tier engine in synchronous Rust (commit 939622d). All 28 Rust tests and 142 Python oracle tests pass.

Implemented 13-language heuristic classifier and safety tier engine in synchronous Rust (commit 939622d). All 28 Rust tests and 142 Python oracle tests pass.

jalmari referenced this issue from a commit

2026-08-30 19:12:53 +03:00

feat(classifier): implement 13-language heuristic engine and safety tiers (#46)