Procure models, tools, and agents through one governed policy layer.
Creation Skills helps procurement, legal, risk, IT, brand, and creative operations teams manage policy documents, approval rules, vendor checks, model permissions, tool limits, agent authority, runtime folders, and evidence requirements before modern media systems scale across a large enterprise.
Approved media agents start as governed folders.
C/Skills packages each approved media agent as a simple folder: instructions, agent rules, brand context, policy context, rights rules, NILP® context, approved skills, tools, procurement records, and evidence checks. It is the three-second mental model that sits in front of the deeper dashboard.
approved-media-agent/ instructions.md agent-rules.md brand-context.md policy-context.md rights-context.md nilp-context.md skills/ media-analysis, rights-review, release-readiness tools/ approved-tools, permission-wrapper procurement/ agent-intake, authority-record, risk-index evidence/ audit-requirements, release-proof-check
C/Skills does not replace agent frameworks. It gives enterprise teams the operating kit that tells approved media agents what they can do, what tools they can use, what evidence they must create, and when a human must approve the action.
One view for policy, procurement, agent identity, runtime readiness, and rollout.
The dashboard is a client-owned operating model. This page explains the model; the backend product is a separate build phase.
Policy Library
One library for governance, procurement, rights, data, agent behaviour, escalation, and audit policies, each with a clear owner, version, and review cadence.
Procurement Readiness
A buyer-friendly view of vendor, contract, security, and data-processing status, with approved and blocked use cases and renewal dates.
Agent Procurement
Authority source, human sponsor, autonomy level, tool tier, reversibility, sub-agent permission, and external system access for goal-driven agents.
Agent Identity Matrix
A required identity record for every approved media agent, agent chain, and sub-agent, so no automation runs anonymously.
Agent Policy Risk Index
A practical operating signal that shows whether an agent is ready for internal pilot, controlled review, restricted use, or held for approval.
Agent Runtime Pack
Runtime-ready folders that package approved operating files: instructions, tools, sandbox rules, channels, connections, schedules, and evals.
Model & Tool Register
Approved models, restricted tools, API access levels, allowed media use cases, and review cadence in one register.
Approval & Evidence Gates
Human approval requirements, evidence checklists, metadata requirements, release proof status, and escalation paths.
Risk & Coverage Readiness
Policy gaps, evidence gaps, first-party and third-party exposure, and Creation Rights rollout readiness before systems scale.
A controlled library, organized for procurement and agent procurement.
Policy documents are organized so buying teams can see what is approved, who owns it, which systems it applies to, and what evidence is required before wider rollout.
Core Governance Policies
Non-negotiable rules, hard prohibitions, approval authority, and enterprise risk appetite.
Procurement & Vendor Policies
How models, tools, vendors, platforms, APIs, and agent systems are reviewed before enterprise use.
Agent Procurement Policies
Bind agent approval to authority source, human sponsor, autonomy level, tool tier, external scope, and evidence capture.
IP Rights & NILP® Policies
Subsisting rights, registered rights, NILP® consent and usage, licensing rules, and release restrictions.
Tool-Use Policies
Approved APIs, read-only, read-write, and execute tools, spending limits, and reversibility rules.
Data & Memory Policies
PII handling, retention, cross-border transfer, memory usage, and source material rules.
Agent Behaviour Policies
Autonomy, goal updating, self-modification, reflection limits, sub-agent permissions, and tool routing.
Escalation Policies
Human approval requirements, risk thresholds, uncertainty handling, and stop conditions.
Audit & Evidence Policies
Logging, traceability, versioning, decision records, evidence capture, release proof, and Creation Rights rollout readiness.
Every approved media agent needs a named identity, sponsor, and evidence path.
The identity matrix is required for each approved media agent, agent chain, and sub-agent. It prevents anonymous automation by showing who authorized the agent, who sponsors it, what it may touch, and what evidence must be saved.
| Identity field | Required value | Why it matters |
|---|---|---|
| Agent ID | Stable identifier for the agent or agent chain. | Prevents anonymous automation and supports audit. |
| Human sponsor | Named accountable enterprise owner. | Shows who is responsible for agent use. |
| Authority source | Policy, contract, approval, brief, or delegation. | Shows why the agent is allowed to act. |
| Operating context | Brand, project, room, release, Skill Pack, or business unit. | Keeps the agent bound to approved company context. |
| Tool tier | Read-only, read-write, or execute. | Separates low-risk review from external action. |
| External boundary | Internal only, external API, publishing, payment, communication, or public release. | Shows when an agent can create commercial exposure. |
| Evidence path | Logs, decision records, metadata, Release Manifest, or Creation Rights protected record. | Keeps agent activity provable after review. |
A practical risk index for agent policies before enterprise rollout.
The index is an operating signal for procurement and governance. It helps show whether an agent is ready for internal pilot, controlled review, restricted use, or held for approval.
| Dimension | Lower-risk signal | Higher-risk signal | Hard stop |
|---|---|---|---|
| Authority | Human sponsor and authority source are present. | Sponsor or authority is unclear. | No sponsor or no authority source. |
| Autonomy | Agent suggests or drafts only. | Agent can approve, route, or execute. | Autonomous approval without a human gate. |
| Tool tier | Read-only access. | Read-write or execute access. | Execute access without an escalation policy. |
| Reversibility | Action can be reversed. | Action is conditional or irreversible. | Irreversible action without an approval rule. |
| External boundary | Internal review only. | External API, publishing, payment, or public release. | External action without an approved destination policy. |
| Rights & NILP® | No protected rights or consent issue. | Rights, likeness, voice, persona, or source material exposure. | Rights or NILP® exposure without evidence. |
| Audit & evidence | Logs, versioning, and decision record present. | Evidence incomplete or not mapped to review. | No audit path for agent action. |
The risk index is an operating signal for procurement and governance. It does not replace legal, security, or insurance review.
Runtime-ready folders for approved media agents.
C/Skills does not replace agent frameworks. It packages the approved operating files those frameworks need: instructions, model settings, tool permissions, sandbox rules, channel policies, connection scopes, schedules, evals, and evidence checks.
agent-runtime-pack/ instructions.md runtime-policy.md approved-model.md approved-tools.md tools/ permission wrappers, risk classification, examples sandbox/ sandbox policy, filesystem access, execution limits channels/ slack, teams, web-chat, api channel policies connections/ connection register, credential scope, approved MCP schedules/ policy drift, vendor renewal, evidence, readiness evals/ rights, NILP, brand safety, policy, release readiness
Named owners, evidence, and release proof before anything moves.
Every approval gate names an owner, an approval trigger, the evidence required, the metadata required, release proof status, an escalation path, and Creation Rights readiness. Approval is held when the authority source, human sponsor, required policy file, rights status, NILP® consent, or execute-tier sign-off is missing.
Two stages turn a generated artifact into a governed outcome.
Output governance classifies the generated artifact. Action governance decides what happens next. Creation Skills consumes output classification signals, then governs the action layer through approval rules, escalation, evidence requirements, audit paths, and Creation Rights readiness. The detectors, classifiers, payment rails, notification systems, and runtime actions belong to the host runtime.
Output governance — classify
Classifies the artifact through content filtering, semantic scoring, watermarking, provenance, and similar checks, producing a verdict the release gate can consume. These detectors run in the host runtime; Creation Skills consumes the signal.
Action governance — decide
Decides what happens next: approval rules, escalation, audit logging, and Creation Rights readiness, with an accountable owner on record. This is the layer Creation Skills governs.
A client-owned dashboard, planned as a separate product phase.
The Enterprise Policy Library & Dashboard is designed to become a client-owned backend for large teams. It will track policy ownership, procurement status, approved models and tools, agent identity, runtime folders, approval gates, evidence requirements, and Creation Rights rollout readiness. This public page explains the model. Backend implementation is handled as a separate product phase.
Creation Skills prepares the operating layer. Creation Rights carries the protected record.
Creation Skills prepares the operating layer
- Policy library records
- Procurement records
- Model and tool register
- Agent identity matrix
- Agent policy risk index
- Runtime pack status
- Approval and evidence gates
- Dashboard readiness views
Creation Rights carries the protected media layer
- Protected records
- NILP®
- Rights metadata
- Evidence vault
- Release proof
- Release Manifest
- Protected media object status
Clear boundaries, by design.
- It does not replace agent frameworks, runtimes, or model providers.
- It does not run the detectors, classifiers, payment rails, or notification systems itself.
- It does not provide legal, security, or insurance advice.
- It does not carry the protected media record — that is Creation Rights.
- It is an enterprise-only, client-owned operating layer, not a public template library.
Build the policy and procurement layer before agents scale.
Creation Skills prepares the operating layer and runtime-ready folders, so approved media agents are governed and evidence-ready before they reach production.
Your company flows.
Your Skill Packs.
Ready for Creation Rights rollout.