Loading your workspace…
Good to see you.
Pull requests, the work that needs you, and your connected services.
The work that needs you
View operational detail →Issues snapshot
Reported observationsSources can describe the same work. Counts are separate observations, not a combined issue total.
Maintenance activity
GitHub PR creationPRs opened by the maintenance bot on Dolabra-built GitHub repositories.
Pull requests
Fleet PRs updated in the last 7 daysRecent outcomes
Source-defined windowsAgents
Service health, current work and source-labelled activity.
Attention queue
Jira + GitHubCurrent failures
Unreachable services + latest failed DAG runsServices
Application and batch service health, with independently reported pipeline state.
Reachable means the health endpoint responded. It does not establish that an API request or batch job is currently running.
Skills library
Available skills for your authorized workspace.
Access
Groups decide who at your organization can use the skills Dolabra limited to them. A change reaches people within 15 minutes (about 30 on Slack); an agent answer already running keeps the groups it started with.
Data Pipelines
Production deployments, DAG states and recent failures.
Failure counts cover seven days; the last-run state is separate. Import errors are not yet reported.
MCP Activity
Server health and reported usage across your MCP connections.
Usage
Active users are distinct signed-in humans. session_init counts initialization-tool calls, not user sessions. Human, service and Slack request classes are separate; daily charts show requests.
Reviewers
GitHub bot throughput and configured custom reviewer lanes.
Merged PRs with a recorded bot review on Dolabra-built GitHub repositories. These are not review-event counts.
Custom review lanes
Partner-owned repositories · last 14 daysReview-session status does not establish PR merge or CI status. Full-agent conversations are not reported by this source.
AI Spend
Provider attribution and platform spend, with source coverage preserved.
Per-service spend
– unavailable · $0 observed zero · Partial data: reported spend only| Service | Group | Status | In flight | Needs human | Modal $ | LLM $ | Source |
|---|
Cost attribution · per AI server
Identifiers only, never keysAI spend · 12 completed months
USDCursor, Claude API, OpenAI API, gemini (api keys + agy adc, once ingested), Claude subscription and total. Subscription is stated vendor cost.
Selective infra spend · 12 completed months
astronomer + snowflake + okta + workos = statedModal and GitHub are metered, pre-credit. Includes the total across all six sources.
All tech vendor spend
12 months current → 12 months projectedLatest vendor catalog snapshot: 12 months at current billed rate, followed by 12 projected months with a three-month ramp. This is a catalog model, not historical invoices.
Connect MCP
Use the current setup guide for your workspace.
WebAI
Open a configured WebAI service to start a conversation.
Create A New Skill
Turn a repeatable method into a skill your team can use.
Create A New Agent
Design an agent that handles a task or workflow.
Usage & Sessions
People using MCP and WebAI in your authorized workspace.