The agentic rag, wrung out.
Tools & PluginsSaturday, October 10, 2026
New OpenClaw plugin stores exact quotes with message IDs and dates. Checksum-verified before load; refuses to start if modified. Local, read-only, free, MIT. Install via `openclaw plugins install clawhub:@python1/verified-memory`.
Addresses a real reliability gap in agent memory (hallucinated or drifted citations).
Tools & PluginsOct 25
New Hermes plugin by @thomasscottbeck that preserves exact user words across turns with precise recall, avoiding LLM summarization loss. Screenshots show the interface; GitHub repo published for review. Teknium suggested submitting to the official plugin catalog.
Tools & PluginsOct 11
Major update to the popular power plugin. SkillEx adds quoting safety, plan-level untested flags, debugging stop-after-three, explicit finish lines, and review caps. FenceGuard keeps parallel coding agents inside sandboxes with no credentials/network by default and prevents settings injection. Memory improvements and multi-terminal efficiency.
Tools & PluginsOct 1
@rlaope shipped a same-day thread on oh-my-hermes (https://github.com/rlaope/oh-my-hermes). It sits beside Hermes and does not patch it. Requests are scored into lanes (rename vs deep search), each lane has its own model chain with fallthrough, prompts are tuned per model and kept only while a benchmark says they help. Author claims the same coding tasks on GPT-6 Astra dropped from $4.29 / 23 min to $0.66 / 5 min. Also ships a cost HUD, verification gates, project memory, and 100+ workflow skills. Earlier post cites 3,000+ GitHub stars.
Never miss a night
One feed for everything, or one per agent harness you run. Rebuilt every night, with full credit on every item.
