aLIZELI/belief-merge

Merges two AI conversations without losing the reasoning, laundering trust or guessing at contradictions. Rather than pasting one thread into the other or lossily summarising, it keeps provenance and applies a confluence theorem with a published counterexample to decide what can be merged. Ask "merge my two bilibili conversations into this one" or drive it with /merge; it applies on the next turn with no config file or restart.

dsh-belief-merge merges two AI conversations without losing the reasoning, laundering trust or guessing at contradictions. Existing tools answer the "two conversations about the same system drift apart" problem by pasting one thread into the other (cheap but unbounded, contradictions left to the reader) or by summarising (lossy exactly where it matters - the reasoning, including the dead ends that stop you repeating them); BeliefMerge instead keeps provenance, so incremental merging stays sound and contradictions are surfaced rather than silently resolved. You can ask in plain language ("merge my two bilibili conversations into this one") or drive it by hand with /merge to list candidates, pick them and check what is in effect.

Memory ★ 2
View on GitHub ↗

Install

dsh plugin --profile web add dsh-belief-merge

Copied verbatim from the README's install line (npm form). Full usage, configuration and limitations live in belief-merge/README.md inside the repo; no config file to edit and no restart is required.

Compatibility

A DeepSeek Harness plugin that applies on the next turn with no config file to edit and no restart. It ships a confluence theorem with a counterexample plus a benchmark, and the package is testable offline (npm test, npm run demo, npm run bench:tight). English README.

Details

Alternatives

863683348/dsh-plugin-focus · akslcw/dsh-negative-ledger · askdkc/kiokuko-dsh

More plugins in Memory

Browse more in Memory

Guides for Memory plugins