I build automation pipelines, agent harnesses, defensive security platforms, and local-first systems where humans stay in the loop—inspectable context, explicit gates, blue-team orchestration, and production-shaped backends (FastAPI, PostgreSQL, Docker, observability).

• Problem: Agent workflows lose intent across sessions; untrusted content reaches LLMs; autonomous AI attackers evade traditional defenses; production ops lack inspectable, human-gated context.
• Solution: Guard–Guide–Defend–Build — SCP (input safety), OpenHarness (handoffs + gates), T3MP3ST_BLU3H4T (autonomous blue team), OpenGrimoire (context graph), plus production platform (CaptionPipeline / Platform API).
• Impact: CaptionPipeline (as-of 2025-12): 256+ caption files, 330+ content hours, <1% errors across 9 production feeds (case study); SCP (as-of 2026-06): 16/16 promptfoo tier probes (OWASP LLM01/LLM06); OpenHarness (as-of 2026-06): harness pin-able by commit SHA, autoresearch Tier B 5/5 on foam-pkm + frontend-a2ui skills; T3MP3ST_BLU3H4T (as-of 2026-07): 8 defensive operator archetypes (MITRE D3FEND), 17-technique AI red team detection, SCP/HITL governance gates.

Portfolio metrics SSOT: metrics.json · refreshed generated_at via refresh_metrics.ps1

flowchart TB
  Intent[Human / Operator intent]
  OH[OpenHarness Guide]
  SCP[SCP Guard]
  T3[T3MP3ST_BLU3H4T Defend]
  WT[moltbook_watchtower Watch]
  AF[Arc_Forge Compounding]

  subgraph build [Build]
    MOP[media-ops-platform Platform]
    OG[OpenGrimoire Context]
  end

  Intent --> OH
  OH --> SCP
  SCP -->|gates tools| MOP
  SCP -->|gates tools| OG
  SCP -->|gates agents| T3
  T3 -.->|detect + respond| WT
  WT -.->|observe| SCP
  AF -.->|mirror| OH
  AF -.->|compound| OG

These seven repos are the proof set—harness → safety → defend → watch → platform → context → compounding.

• CaptionPipeline — automated WhisperX → SCC captions across 9 feeds; as-of 2025-12: 93.5%+ success, peaks 121 files/day → portfolio kit
• SCP guardrail (as-of 2026-06) — 16/16 promptfoo injection/reversal probes before LLM context → SCP README § Impact
• T3MP3ST_BLU3H4T (as-of 2026-07) — autonomous blue team platform; 8 defensive operator archetypes mapped to MITRE D3FEND, 17-technique AI red team detection playbook, governance-first (SCP + org-intent + HITL gates); fork of elder-plinius/T3MP3ST → T3MP3ST_BLU3H4T README
• Agent harness eval — as-of 2026-06: Tier B 5/5 on foam-pkm and frontend-a2ui skills → OpenHarness + MiscRepos autoresearch harness

• Agent harnesses, handoffs, and intent-alignment gates
• Autonomous blue team orchestration, AI red team detection, MITRE D3FEND-aligned defensive operators
• Read-only observability and leak/injection analysis for agent networks
• Local-first context graphs and human↔agent alignment workflows
• LLM input safety (inspect, sanitize, contain, quarantine)
• FastAPI platform APIs, SSO/gateway patterns, homelab observability

Project	One line	CI
OpenHarness	Portable harness: context engineering, handoff flow, intent alignment
T3MP3ST_BLU3H4T	Autonomous blue team platform — 8 D3FEND operators, AI red team detection, governance gates
moltbook_watchtower	Read-only observability for agent networks (leak / injection / behavior)	—
media-ops-platform	CaptionPipeline + Platform API — video captions and homelab integration
OpenGrimoire	Local-first context graph and Sync Session alignment workspace
Arc_Forge	Harness mirror + LLM-Wiki compounding in Obsidian
SCP	Secure Contain Protect — MCP guardrail for LLM inputs (OWASP LLM01/LLM06)

Open an issue on any pinned repo for collaboration or questions.