Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
86 changes: 41 additions & 45 deletions charter/dawn-charter-02.md
Original file line number Diff line number Diff line change
@@ -1,86 +1,82 @@
# Discovery of Agents, Workloads, and Named entities (DAWN)

Many distributed processing environments depend on the interaction
between components that do not have pre-configured capability,
location, or reachability relationships. In order for these components
to operate correctly, the components must be able to discover each
other. For complete generality, we call these components "entities".
Entities may be AI agents, software, services, endpoints, tasks,
workloads, etc.

The principle use case of this working group will be focusing on how
an entity (a device or software, including but not limited to an AI
agent) connected to a network can find available AI agents and how to
contact them about their specific capabilities. The entity will need
to discover the agent's specific properties before proceeding: what
type of agent are they, what communication protocols options are
available, what services do they offer, and what information schemes
they support. Additional more detailed metadata may be communicated
as well, such as how many tokens it can support, whether it is
reactive or proactive, whether it depends on a particular model, and
what data it needs as input.

To support these goals, an interoperable and generic discovery
mechanism is needed that builds on existing protocols and tools,
# Discovery of Agents With Names (DAWN)

Agentic AI systems depend upon the interaction between an entity (a
device or software, including but not limited to an AI agent) and an
AI agent. For this to work, the entity must find available AI agents
and learn how to contact them about their specific capabilities.

The focus this working group will be on how an entity can discover an
agent's specific properties before proceeding: what type of agent they
are, what their reachability information is, what communication protocols
options are available, what services do they offer, and what information
schemes they support. Additional more detailed metadata may be communicated
as well, or may be left out of scope for direct capabilities exchange between
entity and AI agent.

To support these goals, the working group will specify an interoperable
and generic discovery mechanism that builds on existing protocols and tools,
benefits from established trust models, supports proven delegation and
federation architectures, and allows organisations to independently
publish discovery information.

## Scope

Discovery in the DAWN context is limited to being able to "find me an
Discovery in the DAWN context is limited to "find me an
agent to interact with" within an organization, within collaborating
organizations or within local networks. This may include
organizations, or within local networks. This may include
communication about the following example attributes:

- What is the agent's type and classification?
- What communication and security protocols does the agent support?
- What is the minimum requirements for information exchange?
- What are the minimum requirements for information exchange?

Where possible, any solutions work will be built in a modular way
using existing IETF protocols that provide support for any needed
communication, authentication and privacy. The WG will consider mDNS
and DNS as potential initial contact mechanisms upon which to build a
discovery protocol, but may examine other solutions as well.
and DNS as potential mechanisms upon which to build a discovery protocol,
but may examine other solutions as well.

Although initially focused on discovery of AI agents, agentic tools,
and agentic skills, the WG should strive to produce results that are
general and reusable within other discovery contexts if possible.

## Out of Scope

Specifically out of scope of the DAWN WG during its initial launch,
includes:
Specifically out of scope of the DAWN WG during its initial phase are:

- Bulk transfer of larger capabilities beyond the minimal
establishment requirements.
- Bulk transfer of more capability information beyond the minimal
requirements to establish communications.
- All communication between entities beyond the initial discovery process.
- Discovery across the wider Internet.
- Agentic AI stable identifiers
- Entity registration within discovery servers
- Discovery of discovery servers.
- How is the interoperable information schema extensible?
- What attestation evidence and results are available, if any, for the agent?
- How the interoperable information schema is made extensible.
- Availability of attestation evidence and results for agents.

Future DAWN or other WG charters may consider taking on these tasks.

## Deliverables

The DAWN working group will work on the following deliverables,
roughly in this order:
The DAWN working group will work on the following deliverables:

- Terminology: A document defining common DAWN terminology for use in
the other DAWN documents (Optionally published as Informational or
included in the Architecture document).
- Terminology: The definition of common DAWN terminology for use in
the other DAWN documents. (May be published as an Informational RFC or
included in the Architecture document.)

- Discovery Architecture: A document describing the problem space,
requirements and the resulting DAWN architecture (Informational).
requirements, and the resulting DAWN architecture. (To be published
as an Informational RFC, but may be held until the Protocol work is
very stable.)

- Use Cases: A document describing broad use case categories
(Optionally published as Informational).
- Use Cases: A document describing broad use case categories.
(May be remain as an Internet-Draft informing other work or may be
published as an Informational RFC.)

- Protocol: A specification defining the protocol semantics or
- Protocol: A specification describing the applicability of existing
protocols and tools, and defining the protocol semantics or
extensions developed to achieve the initial discovery goals within
the Discovery Architecture.
the Discovery Architecture.

## Coordination with Other Working Groups and Organisations

Expand Down