coilyco-flight-deck/otel-a2a-relay
1
0
Fork 0
Code Issues 12 Pull requests Projects Releases Packages Wiki Activity Actions

Build an otel-a2a-relay MCP server #6

New issue
Closed
opened 2026-05-23 20:55:18 +00:00 by coilysiren · 1 comment
coilysiren commented 2026-05-23 20:55:18 +00:00
Owner
Copy link

Originally filed by @coilysiren on 2026-05-06T23:56:36Z - https://github.com/coilysiren/otel-a2a-relay/issues/4

Companion to #3. Where #3 evaluates pointing an existing MCP (Phoenix) at relay-emitted traces, this one is about exposing the relay itself as an MCP server so an agent can introspect live A2A traffic without going through a downstream tracing backend.

Motivation

The relay sits in the wire path between A2A agents. That's a uniquely good vantage point for an agent-driven inspector: live traffic, no sampling, no exporter lag. An MCP server fronting the relay would let Claude Code (or any MCP client) ask questions like "what did agent X just send to agent Y" or "replay the last failed handshake" without bouncing through Phoenix / Tempo / Jaeger.

Scope to evaluate

  • Tool surface: list active sessions, fetch recent spans, filter by agent pair, replay a captured exchange, tail live traffic.
  • Transport: stdio MCP for local-dev pairing with Claude Code; consider streamable-http for the deployed relay.
  • Auth: how does an agent prove it's allowed to read this traffic when the relay is shared.
  • Storage: in-memory ring buffer vs. lean on whatever OTel exporter is already wired up (likely the latter, query through it instead of duplicating storage).
  • Overlap with #3: if Phoenix MCP covers 80% of the read-path queries, this should focus on the relay-native verbs Phoenix can't offer (live tail, replay, in-flight inspection).

Out of scope

  • Write-path tools (modifying or injecting A2A traffic). Read-only first.

Outcome

Decision doc on whether to build it, what the tool surface looks like, and a tracer-bullet PR if the answer is yes.

_Originally filed by @coilysiren on 2026-05-06T23:56:36Z - [https://github.com/coilysiren/otel-a2a-relay/issues/4](https://github.com/coilysiren/otel-a2a-relay/issues/4)_ Companion to #3. Where #3 evaluates pointing an existing MCP (Phoenix) at relay-emitted traces, this one is about exposing the relay itself as an MCP server so an agent can introspect live A2A traffic without going through a downstream tracing backend. ## Motivation The relay sits in the wire path between A2A agents. That's a uniquely good vantage point for an agent-driven inspector: live traffic, no sampling, no exporter lag. An MCP server fronting the relay would let Claude Code (or any MCP client) ask questions like \"what did agent X just send to agent Y\" or \"replay the last failed handshake\" without bouncing through Phoenix / Tempo / Jaeger. ## Scope to evaluate - Tool surface: list active sessions, fetch recent spans, filter by agent pair, replay a captured exchange, tail live traffic. - Transport: stdio MCP for local-dev pairing with Claude Code; consider streamable-http for the deployed relay. - Auth: how does an agent prove it's allowed to read this traffic when the relay is shared. - Storage: in-memory ring buffer vs. lean on whatever OTel exporter is already wired up (likely the latter, query through it instead of duplicating storage). - Overlap with #3: if Phoenix MCP covers 80% of the read-path queries, this should focus on the relay-native verbs Phoenix can't offer (live tail, replay, in-flight inspection). ## Out of scope - Write-path tools (modifying or injecting A2A traffic). Read-only first. ## Outcome Decision doc on whether to build it, what the tool surface looks like, and a tracer-bullet PR if the answer is yes.
coilysiren commented 2026-05-30 05:44:03 +00:00
Author
Owner
Copy link

Iceboxed in the 2026-05-29 backlog burn-down: Speculative build-an-MCP-server play, no current need. Reopen anytime if it becomes real.

Iceboxed in the 2026-05-29 backlog burn-down: Speculative build-an-MCP-server play, no current need. Reopen anytime if it becomes real.
Sign in to join this conversation.
No Branch/Tag specified
Branches Tags
main
dependabot/uv/uv-e1a0e186ab
No results found.
Labels
Clear labels
  
P0
now / blocking / urgent

  
P1
next / important

  
P2
backlog / will do

  
P3
someday / low

  
P4
icebox / frozen / reopen if it becomes real

No labels
P0
P1
P2
P3
P4
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
coilyco-flight-deck/otel-a2a-relay#6
No description provided.