Expert #12: bounded request-classification substrate #27

Merged
lost-rob0t merged 9 commits from rage/12-request-classifier-red into main 2026-08-31 09:18:29 +00:00
lost-rob0t commented 2026-08-31 09:18:13 +00:00 (Migrated from github.com)

Supersedes draft PR #26 only because the connected ready-for-review transition is unavailable. #26 had zero submitted reviews and zero conversation comments, so no review/discussion history is discarded.

Consumes the first bounded implementation slice of #12 under #8.

Canonical ARADR: research/LLM-LOG-RESEARCH-036-request-classification-expert.org.
Implementation evidence: evidence/LLM-LOG-EVIDENCE-017-request-classifier-red-green.org.

Executed RED: exact head f5e15c2f1fbd8800d767ab8eb701f353ab151668, hosted run 33371681379.
Exact current GREEN: 7d89fffd1158e1d7816fdb09010c15b2a39e629d, hosted run 33376862930; Common Lisp/Tek9/SWI-Prolog, Python expert-policy, and Home Manager gates all passed.

Architecture remains Python capture/transport -> Common Lisp expert runtime -> bounded Tek9 + SWI-Prolog. Rules live in SWI-Prolog; Common Lisp validates typed inputs/results and persists exact source plus derived assertions. No Python rule engine, arbitrary Prolog callable term, whole-corpus materialization, authority escalation, Tek9 repository mutation, or dotfiles consumer mutation is included.

This does not close #12: explicit ambiguity coverage and bounded indexed query acceptance by request/user-message/provider/model/time remain follow-up #12 work.

main currently reports protected=false, required status checks off, and repository rulesets are empty. This PR has no inherited review requirement; zero reviews must not be described as approval. PR #5 remains separate.

Supersedes draft PR #26 only because the connected ready-for-review transition is unavailable. #26 had zero submitted reviews and zero conversation comments, so no review/discussion history is discarded. Consumes the first bounded implementation slice of #12 under #8. Canonical ARADR: `research/LLM-LOG-RESEARCH-036-request-classification-expert.org`. Implementation evidence: `evidence/LLM-LOG-EVIDENCE-017-request-classifier-red-green.org`. Executed RED: exact head `f5e15c2f1fbd8800d767ab8eb701f353ab151668`, hosted run 33371681379. Exact current GREEN: `7d89fffd1158e1d7816fdb09010c15b2a39e629d`, hosted run 33376862930; Common Lisp/Tek9/SWI-Prolog, Python expert-policy, and Home Manager gates all passed. Architecture remains Python capture/transport -> Common Lisp expert runtime -> bounded Tek9 + SWI-Prolog. Rules live in SWI-Prolog; Common Lisp validates typed inputs/results and persists exact source plus derived assertions. No Python rule engine, arbitrary Prolog callable term, whole-corpus materialization, authority escalation, Tek9 repository mutation, or dotfiles consumer mutation is included. This does not close #12: explicit ambiguity coverage and bounded indexed query acceptance by request/user-message/provider/model/time remain follow-up #12 work. `main` currently reports `protected=false`, required status checks off, and repository rulesets are empty. This PR has no inherited review requirement; zero reviews must not be described as approval. PR #5 remains separate.
Sign in to join this conversation.
No description provided.