Open Source

The repos worth your time, grouped. Two of them ship a VALIDATION.md that separates what is verified locally, what CI proves, and what still needs live infrastructure, because a capability nobody has observed is not a capability.

Everything else is at github.com/sheldon904.

Evaluation environments

Task environments with hidden verification. Built because the interesting question is not whether an agent looks right, it is whether your scorer is harder to fool than your agent.

claims-audit-agent

Python and TypeScript

A medical-claims auditing task environment where the labels are verified by an independent engine before any record reaches disk, and the same agent is scored across three orchestration frameworks.

I did not trust labeled data I had not generated, so I built the scorer first and made the agents swappable arms behind it.

Deterministic rule engine pinned at F1 1.000 in CI. Extended thinking cut LangGraph recall from 1.000 to 0.600 and raised Vercel precision from 0.714 to 0.938.

A controlled ablation of stateless RAG, knowledge graph, and hybrid retrieval on one frozen corpus behind one shared typed interface.

Every retrieval comparison I could find changed the corpus along with the architecture, which makes the result unreadable.

Graph-only retrieval hit 98.7% Recall@5 and refused 100% of adversarial distractors where hybrid refused 10%. Two of four pre-registered hypotheses refuted.

agent-eval-methodology

Methodology writeup

The evaluation harness and evidence discipline behind a 14-run pilot of a multi-agent system on a contractor's live operations data.

The methodology is the part worth publishing. The capability number means nothing without the holdout design that produced it.

Two-database holdout, pre-registered thresholds, Cohen's kappa 0.95, 0 fabrications across 51 held-out decisions, one metric marked instrument-invalidated.

cs-intel

Python

A customer-success pipeline combining classical ML, LLM labeling, causal inference, and an approval-gated agent, running fully offline on synthetic data with a known churn process.

A confounder planted on purpose is the only way to check whether your causal correction actually corrects anything.

DoWhy adjustment moves adoption's effect from a naive 30.4 pp to the true 11.7 pp, removing about 88% of the confounding bias. 0.806 ROC-AUC, 55 pytest tests.

MCP servers

Tools for agents where the safety property is structural. An agent cannot misuse a verb that was never exposed.

sparql-mcp

TypeScript

A SPARQL 1.1 MCP server exposing four read tools to agents with no UPDATE surface at all. INSERT, DELETE, DROP, CLEAR, CREATE GRAPH, and LOAD are rejected before they reach the store.

Prompting a model not to write to your knowledge graph is a request. Removing the verb is a guarantee. Host-guarded to localhost, MIT licensed.

Six causal-inference tools over two surfaces (a FastAPI HTTP API and a FastMCP SSE server), wrapping DoWhy, PyMC NUTS with R-hat and ESS diagnostics, EconML CausalForestDML, and causal-learn's PC algorithm behind one uniform envelope.

An agent handed a causal estimate from a cohort of four will use it. This returns a structured "insufficient" below a minimum cohort size instead of a number nobody should trust. Every endpoint but one runs fully offline.

Graph, vector, and holographic memory for agents in one SQLite store, with a Hofstadter-inspired analogy slot, emergent chunking, and usage-driven forgetting.

Most agent memory retrieves what is similar. I wanted one that also retrieves what is analogous: structurally alike but different on the surface.

17-day in-vivo characterization: 1,976 recall events across 235 sessions at 37.4 ms median, zero hot-path API calls.

Infrastructure

Getting an agent fleet onto hardware someone else has to operate, with the trust boundaries written down where a platform team can read them.

backbone-on-k8s

TypeScript and Helm

Migration of my always-on agent runtime from systemd on a droplet to k3s: Cilium CNI, sealed-secrets with no cloud-provider dependency, Keycloak plus oauth2-proxy for SSO, Helm charts, and Grafana.

MCP servers were forked child processes, so the trust boundaries lived implicitly in a process tree. Converting them to network-accessible HTTP services lets NetworkPolicy enforce those boundaries in YAML instead.

VALIDATION.md separates what is verified locally, in CI, and on a live cluster. Recall 3.27 ms with zero probe failures, a 22 MB WAL-mode database backed up online in 0.18 s, NetworkPolicy enforcement confirmed under Cilium, 21 of 21 parity constraints, and $0.00767 per task across 1,111 sessions.

agent-fleet-on-eks

Terraform, Helm, Go, Python

Terraform, Helm, and GitHub Actions shipping agent services to Amazon EKS: test, build, push to ECR, approval gate, Helm upgrade.

Handing a system to a platform team means no long-lived credentials anywhere in CI. GitHub OIDC federation to AWS and IRSA in-cluster, with GitHub Environments and required reviewers gating production.

End-to-end trust flow proven locally with real sockets and HMAC verification. VALIDATION.md marks exactly which rows still need a live AWS account.

Systems

Work outside the agent stack, kept here because it is checkable.

vigil

Java 21 and Spring Boot

A security telemetry system in three services: HMAC-validating ingest to Kafka, in-memory windowed rule detection, and an alert API persisting to PostgreSQL and indexing into Elasticsearch with Kibana dashboards.

Detection thresholds belong in configuration, not in code. All of them live in application.yml.

Brute force at 5+ auth failures from one source in 60 s (HIGH), port scan at 15+ distinct ports in 30 s (MEDIUM), impossible travel across regions within 5 minutes (HIGH). Testcontainers integration tests against real Kafka and Postgres.

Airport

Python and TypeScript

An AI real-estate transaction coordinator: six agents plus Claude Vision document extraction over a Florida compliance engine.

Florida statutory deadlines run in business days from a triggering event, and a miscount has legal consequences, so the deadline math is its own tested module.

15+ statutory requirements covered, 47 tests passing.