zer0dex is a local dual-layer memory pattern for AI agents: a compressed, human-readable markdown index plus a vector store queried automatically before each message. Built for cross-project recall and cross-reference where flat memory files or vector-only RAG fall short. Local-first, low-latency. Reference implementation by Hermes Labs.

lintlang is a static linter for AI agent configs, tool descriptions, and system prompts that runs zero-LLM quality gating in CI. Catches language-level failures (vague tool descriptions, missing stop conditions, schema gaps) before they reach runtime, with deterministic regex + structural detectors and no model calls.

Context-compensation scaffold for LLM evaluation prompts. A short language prefix you prepend so the model discloses prior exposure, scores on quoted evidence only, and hedges on thin evidence — for scorers that can see your CLAUDE.md, memory, or session context. Backend-agnostic. Experimental: variance-reduction effect not yet measured.

hermeneutic is an evidence-first drift gate for AI agents. It mines corrections from your AI chat logs (prior response, user correction, repair), classifies the drift, and runs a cheap-to-expensive pre-flight gate on the next response before drift ships. Regex, then structured scoring, then a pressure probe. MIT, zero dependencies, by Hermes Labs.

forgetted is a Python library for selective memory governance in AI agents: a context-managed window where the agent keeps full read access but its writes to memory files, session logs, deliverables, and (optionally) a vector store silently vanish and are cleaned up on exit. Mid-conversation incognito for agents, one with-block. By Hermes Labs.

Evidence-first structured scoring for AI artifacts: synthesize a rubric, collect citations, score only against quoted evidence, hedge on thin evidence. Every dimension ties to a file:line or quote, with reproducibility receipts. For papers, PRs, prompts, and emails where a defensible LLM-backed score beats a headline number. 7 backends.

Drift-prevention session-init convention card for fresh Claude Code sessions. Injects a self-contained card so a new session opens with its grounding triggers, calibration rules, and tool-map in scope — instead of re-deriving them at minute 30. Per-project, marker-anchored, idempotent, reversible. Bash installer plus MCP server.

fidelis is zero-LLM agent memory for Claude Code and AI agents: a local-first memory layer whose default retrieval path uses BM25, dense vectors, and reciprocal rank fusion with no LLM call. It returns your original passages verbatim instead of paraphrasing and runs fully local. Benchmarked on LongMemEval-S. MIT, by Hermes Labs.

langquant (LPCI) is a scaffold-as-state research artifact testing whether a refreshing language scaffold can serve as the sole working state for a stateless LLM. In one A/B run (n=1/condition, 20 turns) the model held coherence with zero history; transfer entropy dropped 0.608 to 0.085, a large reduction, not zero. Single observation, not a proof.