Protocol
AdCP defines the shape of the transaction.
Campaign briefs, inventory requests, revisions, and execution messages become predictable objects instead of fragile one-off payloads.
AdSkills
Protocol clarity, skill-level intelligence.
AdCP + AdSkills
The protocol is the contract. The skill is the operator.
AdCP does not compete with AdSkills. It gives them a stable operating surface. One defines how advertising systems exchange work; the other defines how intelligent advertising work gets done.
Skill
AdSkills supply the advertising judgement.
The skill interprets those objects, applies domain logic, and chooses what should happen next based on performance, constraints, and context.
Result
Agents can act without improvising the entire workflow.
The protocol keeps systems aligned; the skill keeps the decision-making credible. That separation is what makes agentic workflows repeatable.
Scenario
What this looks like in an actual campaign workflow.
The cleanest way to understand the split is to follow one planning task from intake to execution.
01
A campaign brief arrives in AdCP format
The brief includes market, objective, budget, time horizon, and channel constraints in a shared schema every compliant system can read.
02
A planning skill interprets the brief
The skill evaluates audience fit, channel tradeoffs, historical learnings, and pacing logic without requiring the agent to rediscover advertising fundamentals.
03
Recommendations are returned as structured actions
Media allocations, rationale, and follow-up transactions can be handed back through AdCP instead of buried in natural-language output.
04
Execution stays portable
Because the handoff remains protocol-native, another system or skill can review, approve, or execute the plan without translation work.
Why it matters
Protocol-first systems get much more useful when the expertise layer is separate and reusable.
Standardization without flattening judgement
AdCP standardizes the language. Skills preserve the expertise.
Portable automation
Workflows are less dependent on one vendor because the data contracts and the intelligence layer are both modular.
Cleaner human oversight
Teams can inspect both the transaction format and the decision logic instead of trusting a hidden chain of prompts.