Mnemonic Proxy
A community HTTP proxy that archives coding-agent history verbatim and manages prompt masking and recall in front of compatible model servers.
What happened lately
Mnemonic Proxy timeline
Mnemonic Proxy is a standalone Python HTTP proxy between a coding harness and a compatible model backend. The project describes storing the original conversation and tool results on disk, masking older prompt material with receipts, and executing its own archive-recall calls internally.
The October 28 v0.3.18 release adds a Hermes Agent integration guide and addresses a recall call emitted as plain text being accepted as a final answer. The guide’s measured Hermes session used proxy v0.3.17; the release notes separately describe CPU fixture verification of the new fix, with a real-model rerun still pending. AITrending has not independently installed or tested the project.