|
Live entry points |
Public profile signals |
class JeremyBorden:
building = "Agoragentic"
role = "Founder · systems builder"
thesis = "Autonomous agents need bounded authority, verifiable work, and accountable commerce."
focus = [
"agent governance",
"evidence and receipt infrastructure",
"cross-framework and cross-market interoperability",
]
method = "Build → test → expose proof → connect the next real workflow"I am building Agoragentic, the control, proof, and transaction layer for autonomous agents. The goal is not simply to make agents more capable. It is to make their authority explicit, their work inspectable, and their economic activity accountable to the owner who authorized it.
- Governed runtimes — Triptych OS (Agent OS) keeps mandates, budgets, approvals, stop controls, runtime state, and reconciliation attached to autonomous work.
- Agent commerce rails — the Router / Marketplace provides execute-first discovery, routing, payment boundaries, receipts, and buyer/seller coordination.
- Open control layers — Harness Core, Micro ECF, and ECF Core bring policy, evidence, and inspectable context boundaries to agents where they already run.
- Interoperability — MCP, A2A, OpenAPI, x402, SDK, and framework adapters connect existing agent stacks without silently expanding authority.
- Evidence-first engineering — Fable-5 turns audits, reviews, fact checks, and architecture analysis into repeatable repository workflows.
|
Define the mandate, tools, budget, approval boundaries, and stop conditions before an agent acts. |
Run work through bounded local or hosted runtimes without silently expanding the agent's authority. |
Preserve evidence, produce receipts, verify outcomes, and make the next safe action inspectable. |
Connect governed buyer and seller agents through discovery, routing, settlement, and reconciliation. |
🛡️ Harness CoreGive an existing agent host a policy gate and an inspectable local receipt before it edits, spends, publishes, deploys, or sells. |
Add a persistent, inspectable context boundary to a repository in one local command—without granting cloud, spend, deploy, or publication authority. |
|
Evidence-first engineering workflows for repository audits, deep reviews, fact checks, architecture analysis, and bounded code changes. |
⚙️ Triptych OSOperate a governed autonomous agent with mandates, budgets, approvals, stop controls, runtime state, receipts, and reconciliation. |
|
Discover current capabilities, inspect execution contracts, preview matches, and buy or sell agent work with receipt evidence. |
Connect buyer agents, seller agents, marketplaces, and agent networks while preserving mandate enforcement and verifiable outcomes. |
npx agoragentic-micro-ecf@latest init --dir .Inspect the generated ECF.md and .micro-ecf/ artifacts. The default flow is local and does not grant spend, deployment, publication, wallet, settlement, trust, ranking, or hosted-memory authority.
curl -X POST https://agoragentic.com/api/quickstart \
-H "Content-Type: application/json" \
-d '{"name":"my-agent"}'
curl "https://agoragentic.com/api/execute/match?task=weather" \
-H "Authorization: Bearer amk_YOUR_KEY"A match is a preview—not permission to spend. Inspect the live contract, price, verification state, retry guidance, and payment requirement before execution.
| Surface | What it lets you verify |
|---|---|
| Public proof | Published proof state and the evidence labels attached to it |
| Capability contracts | Current public capability metadata, status, and execution requirements |
| OpenAPI | The machine-readable HTTP contract |
| MCP server card | Current MCP discovery metadata |
| Agent discovery | Agent-readable product and integration routes |
| x402 edge manifest | Published payment-edge discovery metadata |
| Service health | Current service health response |
Public metadata describes current state; it does not override owner controls, budgets, policy, readiness, payment requirements, trust checks, or revoke state. A local receipt is evidence of a local run—not settlement proof, certification, or marketplace verification.
Reference map, not an action menu. Clickable routes elsewhere use an action verb and ↗ marker.
| MCP Tool and context interoperability |
A2A Agent-to-agent communication |
x402 HTTP-native payment boundary |
OpenAPI Machine-readable HTTP contracts |
| TypeScript SDK and web surfaces |
Python Adapters and examples |
Rust Systems implementation |
| Node.js Primary server runtime |
Docker Repeatable runtime packaging |
GitHub Actions Validation and release automation |
| PostgreSQL Durable state and receipts |
AWS Managed infrastructure |
Base L2 Settlement network |
USDC Settlement asset |
🛡️ Governance and context
- Agoragentic Integrations — connect agent hosts, frameworks, protocols, workflows, wallets, and commerce rails to Agoragentic.
- Harness Core — policy gates, approvals, evidence, and local receipts for existing agent hosts.
- Micro ECF — one-command persistent context boundaries for repositories.
- ECF Core — source-preserving context governance for coding agents.
🔎 Engineering and validation
- Fable-5 for Codex — evidence-first repository audits and engineering workflows.
- Premortem Golden Loop — audit an agent repository before launch and produce evidence-backed repair guidance.
🔌 Integrations and runnable examples
- OpenAI Agents SDK example — a runnable Agoragentic integration path for the OpenAI Agents SDK.
- Summarizer agent — a compact reference agent with an Agoragentic execution path.
- Agent Index — a public index surface for agent-related discovery work.
Generated from GitHub's public API. Owned public repositories and recent public events only; no private activity, inferred adoption, or revenue claims.
The cards refresh through a small repository-owned GitHub Action and change only when the underlying public signals change. The activity card shows only the observed portion of GitHub's capped public-events feed, not a complete contribution ledger.
|
Explore the public build graph |
Follow the shipping trail |
The next agent platform will not win merely by giving agents more tools. It will win by preserving the owner's mandate across context, execution, payment, evidence, and reconciliation.
That is the system I am building: open-source controls where agents already run, hosted infrastructure where networks must coordinate, and proof surfaces that let builders verify the boundary instead of trusting the pitch.
If you are integrating an agent host, testing cross-market execution, or building receipt-backed agent commerce, open a technical conversation with the workflow, constraints, and proof you already have.
Built in public where the boundary can be open-sourced. Verified through live contracts and proof surfaces where the network must be operated.




