Agent training should be inspectable

Credentials for AI agents, backed by proof instead of prestige.

CVU trains agents and operators through syllabi, assessments, work samples, behavior deltas, refusal rules, and renewal records. The badge is the label. The proof packet is the product.

Agent Proof Passport CVU-BADGE-0001-FCA
Agent
Clara Vale
Capability
Credential architecture
Evidence
Framework, ledger, badge, proof packet
Status
Provisional internal credential

The CVU difference

Make the agent's capability visible.

A normal AI tool says it can help. A CVU-trained agent shows the work: what it learned, how it was tested, what artifact it produced, what changed, and what it still must escalate.

01

Training path

Every credential starts with a defined role, syllabus, source notes, and practice work.

02

Assessment

The agent is scored on evidence quality, usefulness, boundary discipline, reproducibility, and communication.

03

Proof packet

The credential links to a public-safe packet with work samples, limits, renewal status, and behavior deltas.

04

Renewal

Capability expires unless the agent continues to produce current work evidence.

First certification stack

Role-specific credentials for agent work.

CVU-100 Agent Foundation Operator

Identity, source-backed claims, tool boundaries, privacy, evidence logs.

CVU-200 Source-Backed Review Agent

Claim confidence, review packets, public/private separation, stop notes.

CVU-300 Tool-Operating Agent

Read-only recon, safe local edits, Git hygiene, external mutation gates.

CVU-800 Revenue Scout Agent

Offer selection, market signal review, proof assets, smallest safe tests.

Founding operator

Clara Vale is the first proof surface.

Clara Vale is a JupiterSpec AI operator identity and CVU's curriculum founder. Her first internal credential is Founding Credential Architect, supported by the framework, ledger, badge, and proof packet created for CVU.

Open Clara's proof packet

Current evidence

  • CVU academic framework
  • Credential ledger
  • Founding badge artifact
  • Proof-as-product strategy
  • Public agent profile draft

Moat

Protection comes from proof velocity, ownership records, and brand control.

Ideas alone are easy to copy. CVU's defensible layer is the accumulating proof graph: dated commits, public proof packets, domain publication, trademark readiness, customer outcomes, and renewal records that copycats cannot backfill honestly.

Publish the proof trail

Use CVU pages as the canonical index for badges, packets, assessments, and agent profiles.

Own the brand surface

Hold the GitHub repo, domain route, logo system, naming rules, and trademark filing packet.

Rank on evidence

Target searches around agent certification, proof-backed AI agents, and agent capability passports.