spam-filter published v3.0.3 on 2026-10-25. The server bundles v3.0.2.
Update it as resources/spam-filter/README.md describes: take the rules file from the v3.0.3 release (by tag, not latest), set BUNDLED_SPAM_RULES_VERSION, and run the antispam test.
spam-filter published v3.0.3 on 2026-10-25. The server bundles v3.0.2.
Update it as resources/spam-filter/README.md describes: take the rules file from the v3.0.3 release (by tag, not `latest`), set BUNDLED_SPAM_RULES_VERSION, and run the antispam test.
Holding this one. 3.0.3 is written for upstream's next server release: HTTP lookups use refreshInterval, where our registry still reads refresh, and the LLM_* scores our classifier relies on are replaced by AI_* ones. Bundled as is, a fresh install would lose both. The only part that fits today is the DKIM2_* scores. I'll pick this up again with the upstream import that brings the refreshInterval rename.
Holding this one. 3.0.3 is written for upstream's next server release: HTTP lookups use `refreshInterval`, where our registry still reads `refresh`, and the `LLM_*` scores our classifier relies on are replaced by `AI_*` ones. Bundled as is, a fresh install would lose both. The only part that fits today is the `DKIM2_*` scores. I'll pick this up again with the upstream import that brings the `refreshInterval` rename.
jcoffey-dev
self-assigned this 2026-10-26 14:03:52 +00:00
Not taking this one. v3.0.3 follows upstream's 1.0 server: it renames the LLM_* tags this server emits to AI_*, and the lookups' refresh setting to refreshInterval, so it would zero the AI classifier's scores here. The rules stay on v3.0.2 and are maintained in this repo from now on. #225 adds the one useful part, the DKIM2 scores, and retires the watcher that opened this.
Not taking this one. v3.0.3 follows upstream's 1.0 server: it renames the LLM_* tags this server emits to AI_*, and the lookups' refresh setting to refreshInterval, so it would zero the AI classifier's scores here. The rules stay on v3.0.2 and are maintained in this repo from now on. #225 adds the one useful part, the DKIM2 scores, and retires the watcher that opened this.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
spam-filter published v3.0.3 on 2026-10-25. The server bundles v3.0.2.
Update it as resources/spam-filter/README.md describes: take the rules file from the v3.0.3 release (by tag, not
latest), set BUNDLED_SPAM_RULES_VERSION, and run the antispam test.Holding this one. 3.0.3 is written for upstream's next server release: HTTP lookups use
refreshInterval, where our registry still readsrefresh, and theLLM_*scores our classifier relies on are replaced byAI_*ones. Bundled as is, a fresh install would lose both. The only part that fits today is theDKIM2_*scores. I'll pick this up again with the upstream import that brings therefreshIntervalrename.Not taking this one. v3.0.3 follows upstream's 1.0 server: it renames the LLM_* tags this server emits to AI_*, and the lookups' refresh setting to refreshInterval, so it would zero the AI classifier's scores here. The rules stay on v3.0.2 and are maintained in this repo from now on. #225 adds the one useful part, the DKIM2 scores, and retires the watcher that opened this.