Boot Quadranscustom agentic harness
Quadrans custom agentic harness

Rotate your phone to landscape to unlock the full animated presentation.

AI writes code faster than your team can trust. scroll AI agent emitting output stream · quantity, not quality Duplication and verbosity 1 Hallucinations 2 Unreviewed handoff 3 Non-functional failures 4 Missing acceptance tests 5 five failure classes Human reviewer single-threaded burnout context: overloaded Our Solution scroll
1 · The problem

AI writes code faster than your team can trust.

An AI agent emits a stream of output: fast, prolific, and mostly wrong. Five failure classes keep showing up: duplication and verbosity, hallucinations, unreviewed handoff, non-functional failures, missing acceptance tests. All of it lands on a single engineer who burns out trying to keep up.

The sources of value worth protecting are your attention, your output correctness, and your action safety. Without a harness, AI generates faster than your team can trust.

2 The System
A harness around the agent. Task input 1 Context 2 Workflows 3 Verification 4 Human review valuable outcome Knowledge base compounding loop-back decision-742 doc-318 precedent-95 decision-431 doc-657 precedent-203 two-way knowledge transfer skills integrations tools Verification every output verified to knowledge base nothing stays in the session
2 · The system

A harness around the agent.

The solution wraps the agent in a complete pipeline: context feeds the agent, workflows drive execution, verification checks every output, and user review gates the final decision. A knowledge base compounds underneath, feeding every session and drawing from it.

Pluggable modules handle the edges: skills, integrations, tools. Verification sits in the middle because every output gets checked before it reaches anyone. Users stay in the loop at every gate. Nothing stays in the session.

3 Close-up · Context engineering
context · 1:1 Context compounds. AGENTS.md entrypoint Session live context Knowledge base compounding External context catalog pluggable sources GitHub Figma Notion CI consolidate context learning loop every session makes the next one better sessions Fig. 3 · context engineering and compounding loop: session → consolidate → KB → context
3 · Context engineering

Context engineering

AGENTS.md is where context discovery starts, the entrypoint that points the agent to your knowledge base, your session, and external sources. A learning loop captures what happened in each session, consolidates it, and feeds it back as context for the next one. The knowledge base compounds: decisions, precedents, docs, all stacking up.

External sources plug in through a catalog: GitHub, Figma, Notion, CI, so the agent doesn't work in a vacuum. Every session makes the next one better. Nothing gets lost between sessions; everything becomes part of the living system.

4 Close-up · Workflows & parallelization
Work is a pipeline, not a chat. requirement intent 1 Intake KB research 2 Spec design Delta spec 3 Architecture, infra, UX ADRs 4 Implementation plan Task plan 5 Coding subagent 1 subagent 2 subagent 3 parallel subagents in isolated worktrees 6 Agent-led verification current merged Test report 7 Assisted human review 8 Deployment and observation 9 Knowledge capture Precedents runtime hooks enforce compliance custom Skills drive work hand-off checkpoints for humans checkpoint checkpoint checkpoint checkpoint
4 · Workflows

Work is a pipeline, not a chat.

Nine steps: intake, spec design, architecture and infrastructure, implementation plan, coding, agent-led verification, assisted user review, deployment and observation, knowledge capture. Custom skills drive each step. Runtime hooks enforce compliance. Hand-off checkpoints appear at the right time where users take over.

At step 5, the pipeline fans out to parallel subagents working in isolated git worktrees, that's where the speed comes from without sacrificing correctness. The agent runs its own verification next. Knowledge capture feeds everything back to intake. The knowledge base keeps compounding.

5 Close-up · Human sensemaking
Two-way knowledge transfer. Agent autonomous execution Human judgment · intent true cost of speed: understanding 5 review queue PR #118 PR #203 PR #311 PR #427 PR #509 agent → human · walkthroughs, visual explainers human → agent · intent, tradeoffs, values decision.md adr-7 spec.md tradeoffs.md walkthrough.html decision records Adversarial review agent PR #214 attacks the diff before humans read it PR #509 1 anchored
5 · Human sensemaking

Two-way knowledge transfer.

The agent offers walkthroughs and visual explainers. The engineer explains intent, tradeoffs, and values. Five artifacts anchor every action: decision records, ADRs, specs, tradeoff docs, walkthroughs, nothing lives only in the session.

An adversarial review agent attacks the diff before engineers read it, catching assumptions, oversights, the failure modes engineers would miss. The review queue collapses from five PRs to one. Understanding is the true cost of speed, and this engine pays it.

6 Close-up · Verification
Vibing not allowed. spec-driven verification Spec the contract requirements · acceptance Diff 6 files changed +412 −187 verification pipeline Claims agent · verify $ quadrans verify --spec ./spec writing 14 assertions … ok executing suite … pass 14 · fail 0 proof Security threat scan Quality lint · tests Compliance policy map Frameworks GDPR PCI HIPAA SOC 1 and 2 Custom fail any gate Blocked back to coding Verified 14 of 14 checks · proof attached every change carries its proof
6 · Verification

Vibing not allowed.

The spec is the contract. The diff is the change. The agent writes the test suite, executes it, and produces proof. Then every output passes through a gate chain: security analysis, quality checks, and compliance verification, GDPR, PCI, HIPAA, SOC 1 and 2, custom frameworks.

Fail any gate and it's blocked, back to coding. Pass all of them and you get verified: proof attached, every change carrying its evidence. This is how AI becomes safe for mission-critical software.

7 Pluggable Modules
The core may shift. The edges are yours. pluggable modules Codex Claude Code Pi skills integrations Onboarding Workflow steps KB maintenance Learnings TDD and SDV Parallel agents Tasks Docs Design Git CI and CD Observability PR lifecycle Pipeline status deterministic work moved to code model- and runtime-portable encodes team standards
7 · Pluggable modules

The core stays. The edges are yours.

The core harness stays fixed, skills, integrations, and custom tools all plug in through sockets. Skills handle onboarding, workflow steps, knowledge base maintenance, learnings, TDD, parallel agents. Integrations connect to tasks, docs, design, Git, CI/CD, and observability. Custom tools handle deterministic work like PR lifecycle and pipeline status.

The whole system is model- and runtime-portable, encoding team standards without locking you to any vendor. As the model landscape shifts, your edges stay yours. Bad practices get prevented by the structure itself.