Datadog Developers

The developer side of Datadog. Service status:

[Amy Zhou @amymzhou]

ppl said my last video was not cringe, and I'm choosing to believe them!

did you know MCP released a new spec (2026-07-28)? I couldn't cover everything, but I break down the implications of some of those changes in this video... (Stateless, Tasks, etc) @datadoghq @datadogdevs

[Greptile @greptile]

Join us: https://t.co/CCaHwGambU

[Datadog Developers @datadogdevs]

@greptile @aiDotEngineer @datadoghq @dakshgup @nerdsane can't wait 🔥

[Greptile @greptile]

In NYC for @aiDotEngineer next week? Join us and @datadoghq for a rooftop happy hour.

Drinks, light bites, and a fireside chat with @dakshgup and @nerdsane on "What changes when AI starts writing the code?"

Tues, Oct 13.

[Datadog Developers @datadogdevs]

@ryanyee123 @datadoghq love this 👏🏻

[Datadog Developers @datadogdevs]

Want to see the New York skyline after-dark while learning about what changes when AI starts writing code?!

Then save your spot for our upcoming @greptile x @datadoghq rooftop happy hour featuring CEO, @dakshgup and @nerdsane!

grab your spot 👉🏻 https://t.co/sfcPDSwa7m

[Datadog Developers @datadogdevs]

Unstructured logs. Searchable fields. No Grok required. 👀

See how Tap to Parse turns values buried inside raw log messages into attributes you can actually filter, group, and investigate.

In this example: payment success drops → isolate the failing provider → keep digging, in about a minute.

Tap to Parse generates the parsing rule for you, and you can review or edit it before applying anything.

See how it works across @datadoghq Log Explorer, Log Pipelines, and Observability Pipelines → https://t.co/gcymx0ZdMe

[Datadog, Inc. @datadoghq]

Your alert fires. Your team is in Slack. Your evidence is everywhere else.

Bits Chat puts Bits AI in the channel where you're already working. Investigate, troubleshoot, ship the fix, write the postmortem. Learn more on the blog: https://t.co/B6DC5nqoxX

[Datadog Developers @datadogdevs]

shipped this week at Datadog 🚀:

1. Agents burning tokens one call at a time? Code Execution mode: run agent code in a sandbox against our APIs. 70% fewer tokens, 20% more accurate.

2. Telemetry outpacing release cycles? RUM Remote Config: change SDK behavior, no redeploy.

3. @Shopify or @salesforce hard to instrument? New RUM + Product Analytics integrations, out of the box.

4. Need OTel traces in @datadoghq without extra infra? OTLP Traces Intake: straight from your exporter.

5. Code issues surfacing post-merge? Bitbucket Cloud integration: scanning + PR comments pre-merge.

6. Kafka issues scattered across tools? Kafka Console: one place to investigate, with MCP support.

7. Incident response bouncing between tools? Bits Chat now native in @SlackHQ: @ Datadog investigate → auto-generated fix PR

Full write-up + docs → https://t.co/d9OlqDYEgO

[Datadog Developers @datadogdevs]

the @datadoghq MCP is doing VIP stuff.

[Amy Zhou @amymzhou]

@georgeelizabeth made me post this 🤡 🙈 is this the beginning of my influencer era?

we're working hard on making the datadog mcp server the best it can be...planning to share learnings/new features more often. what videos would you want to see next? @datadogdevs @datadoghq

[Datadog Developers @datadogdevs]

@amymzhou @georgeelizabeth @datadoghq the MCP VIP 👏🏻

[Amy Zhou @amymzhou]

@georgeelizabeth made me post this 🤡 🙈 is this the beginning of my influencer era?

we're working hard on making the datadog mcp server the best it can be...planning to share learnings/new features more often. what videos would you want to see next? @datadogdevs @datadoghq

[Datadog Developers @datadogdevs]

@bradcypert @datadoghq we approve this message 🫡

[Datadog Developers @datadogdevs]

@bot i spy @datadoghq

[Grok Bot @bot]

Grok Bot is now more powerful for building software.

Bots can hand off coding tasks to Cursor, manage your PRs with GitHub and Origin plugins, and share video demos of what they build.

[Datadog Developers @datadogdevs]

Datadog + @awscloud put together a practical pattern for securing agentic AI in production - combining Amazon Bedrock Guardrails, AgentCore, Strands Agents, and @datadoghq AI Guard.

Key takeaway: securing an agent can’t stop at the prompt.

You need controls around what it can access, what it decides to do, and the tools it calls - with enough context to catch things like indirect prompt injection before an action executes.

give it a read now 👉🏻 https://t.co/smkGyGeUQU

[Datadog Developers @datadogdevs]

Async Python makes traditional flame graphs harder to interpret.

We rebuilt @datadoghq's Python profiler to preserve the relationships between asyncio tasks, and cut profiler overhead by 60%+ in the process.

see how we built it → https://t.co/R8dfIQucR8

[Datadog Developers @datadogdevs]

@kaixin_tai we love to see @maxirodgo

[Datadog Developers @datadogdevs]

@DishaAnadkat @datadoghq great pic 👏🏻