[SUPERSEDED] Parallel DeepSeek Harness frontend backed by Prolog-RLM semantics #124
Labels
No labels
bug
documentation
duplicate
enhancement
good first issue
help wanted
invalid
question
wontfix
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
nsaspy/prolog-rlm#124
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Build a parallel downstream PrologAgent implementation from the official MIT-licensed
deepseek-ai/deepseek-harnesssurface while keepingprolog-rlmauthoritative.Required invariants:
agentPrologrenderer work; do not regressprolog_agent_ui_v1.First slice should establish the pinned upstream boundary, a Prolog bridge protocol, durable settings/session ownership, no-compaction contract, provider routing, tests, and documentation. Follow-up slices can deepen UI coverage and tool/approval presentation.
ADARD checkpoint
Plugin-only DeepSeek Harness core research/design is now recorded in #184: https://github.com/lost-rob0t/prolog-rlm/issues/184
#184 is AWAITING APPROVAL. It narrows the first milestone to the official unmodified Harness Web/headless surfaces + one out-of-tree Prolog-backed
AgentFactoryplugin + a narrow Prolog bridge + Nix integration. It explicitly rejects treating the current broad #125 implementation as automatically approved scope.RAGE must not begin on this redesign until #184 is explicitly approved.
Superseded by the approved plugin-only Harness architecture in #184 and cleanup/cutover work in #186. This older parallel-Harness direction should not be used as implementation authority.