epistemic-engineering

Lean governance for AI-assisted development: requirements tracking, test enforcement, epistemic context compression, auditable evidence, Grace local REPL, and coding-agent integrations.

Certified log-concavity of the Riemann-Jacobi kernel with Arb/FLINT ball-arithmetic certificates. Source-critical audit of the Polya-type real-zero criterion. No unconditional RH claim.

Research and DOI publications hub for Hermes Labs: a Zenodo-canonical, DOI-anchored index of papers on epistemic and hermeneutic failure modes in large language models — null-result asymmetry, source-status bias, silent instruction relaxation. Each entry links its DOI, summary, and the tools that operationalize the findings.

Archival repository for the Zenodo technical note Behavioral Canarying for Prompt Injection: Powerless Model Probes with Explicit Coverage Semantics.

Single-page reference card: eight failure modes of AI-augmented knowledge work, each with countermeasure. Share it, print it, or load it into an agent's context.

Archival repository for the Zenodo technical note Tool Differentia: Relational Static Analysis for AI Agent Tool Descriptions. The check itself ships in hermes-labs-ai/lintlang (H1.6).

Applied hermeneutics, linguistic attractors, and the limits of model self-report in language-model systems.

Archival repository for the Zenodo paper The Asymmetric Burden of Proof.

Archival repository for the Zenodo paper A Taxonomy of Epistemic Failure Modes in Large Language Models.

Measurement validity, construct validity, and unsupported claims derived from AI agent telemetry.

AI reliability engineering for production AI agents and LLM applications. Open-source tools and research that find silent failures in prompts, tools, retrieval, memory, policies, runtime controls, agent evaluation, and evidence.

Self-governing orchestration framework with 8 phases, 4 human review gates, BLAKE3 audit chain, and ethical axiom enforcement