Dashboards

Colony Cloud provides four read-only operator surfaces under the Dashboard group in the left navigation. Each surface shows a different slice of pipeline activity; together they give the operator a complete picture of fleet health, trends, and issues requiring intervention.

Operator Home

Route: /dashboard/home

Operator Home is the first screen after sign-in and the primary daily-use surface. It shows the real-time health of the autonomous fleet at a glance.

Key metrics

Metric	What it shows
In-flight	Issues actively progressing through the pipeline — all non-terminal, non-blocked, non-paused states. Click to drill into Pipeline Live.
Blocked	Issues that cannot progress without external intervention (dependency-blocked, failure-blocked, needs-clarification). Click to open Needs Attention filtered to blocked issues.
Awaiting Review	Issues where Colony has opened a PR and is waiting for a human code review (human-review-ready). Click to open Needs Attention filtered to review-ready issues.
Cost today	LLM token spend by Colony agents since midnight UTC. Click to open the Executive dashboard for historical cost trends.

All four KPI cards refresh on a 3-second polling cadence. Values reflect the live Pipeline Store state — not a delayed snapshot.

Worker and queue status

Below the KPIs, a secondary row shows:

• Workers — count of worker containers currently registered and reporting heartbeats.
• Pending tasks — tasks queued but not yet assigned to a worker. A growing pending count relative to worker count signals that capacity may be undersized.
• DLQ depth — messages in the dead-letter queue (tasks that exhausted all retry attempts). A non-zero DLQ depth always requires operator attention.

Caution

A non-zero DLQ depth means work has permanently failed and is waiting for manual resolution. Open Needs Attention to see which issues are affected and use the slash commands to retry or cancel.

Executive

Route: /dashboard/executive

The Executive dashboard shows performance trends over time. It is the right surface for weekly reviews, sprint retrospectives, and cost tracking conversations with stakeholders.

Use the 7-day / 30-day toggle in the upper right to set the analysis window. All metrics on the page update to reflect the selected window.

Trend metrics

Metric	What it shows
Throughput	Issues completed (PR merged) within the selected window.
Cycle Time	Mean elapsed time from issue creation to PR merge across completed issues. Higher than usual indicates stalled stages — drill into Pipeline Live to find them.
Success Rate	Percentage of completed issues that merged successfully, excluding failed or abandoned attempts.
Review Pass Rate	Fraction of merged PRs that passed code review on the first submission without a changes-requested cycle.
Time to First PR	Median elapsed time from issue creation to the first PR opened.
Avg Review Cycles	Average number of review rounds per PR before approval. More than two cycles consistently signals a configuration or issue-quality problem.
PR Revert Rate	Fraction of merged PRs subsequently reverted. Spikes here warrant a review of the issue quality and review strictness.
Agent Attribution	Share of merged PRs handled by Colony agents versus human contributors.
Cost (windowed)	Total LLM token spend across all Colony agents for the selected window.

Week-over-week comparison

Each trend metric displays a delta badge comparing the current window to the prior equivalent window (e.g., this week vs. last week). Green indicates improvement; amber indicates a meaningful regression worth investigating.

Note

Cycle Time currently uses a mean (AVG) calculation. A future update will switch to median (PERCENTILE_CONT(0.5)) for better outlier resistance. Watch for that change if you use cycle time as a planning input.

Pipeline Live

Route: /dashboard/pipeline-live

Pipeline Live shows the real-time state of every issue in the pipeline as a visual state-machine grid. It is the right surface when you need to understand what is happening right now — not trends, but the current distribution of work across pipeline states.

State grid

Each pipeline state appears as a node in the grid. The count inside each node is the number of issues currently in that state. Click any count to open a filtered issue list scoped to that state.

States are organized into groups:

• Intake — new, analyzing
• Planning — planning, ready-for-dev
• Development — in-development, waiting-for-subtasks
• Review — in-review, human-review-ready, changes-requested
• Merge — merge-pending
• Blocked / Recovery — dependency-blocked, failure-blocked, needs-clarification
• Terminal — done, cancelled, paused

List and board views

Toggle between list view (tabular, sortable by state and age) and board view (column per state, cards per issue) using the view controls in the upper right. Both views update on the same 3-second polling cadence as Operator Home.

Tip

Sort list view by age descending to surface issues that have been in the same state for an unusually long time. Issues that have not transitioned in more than a few hours are candidates for operator intervention.

Needs Attention

Route: /dashboard/attention

Note

The path /attention (without the /dashboard/ prefix) is a permanent redirect to /dashboard/attention. Bookmarks and external links using the short form continue to work.

Needs Attention surfaces issues that require operator action — the subset of the pipeline that the autonomous recovery protocols could not resolve on their own.

Attention categories

Category	What appears here
Blocked	Issues in dependency-blocked, failure-blocked, or needs-clarification that have not progressed after the colony’s automatic retry budget was exhausted.
Stalled	Issues in an active state (e.g., in-development) that have not produced a state transition in longer than the stall threshold (default: 2 hours).
Review loops	Issues that have cycled through changes-requested → in-development → in-review more than three times. This usually signals a problem in .colony/conventions.md or the issue itself, not the individual PR.
DLQ entries	Tasks in the dead-letter queue with a link to the originating issue and the last error message.

Common interventions

From the Needs Attention list, you can act directly on each issue:

Situation	Action
Worker stalled or timed out	Click Retry (equivalent to /colony:retry on the issue).
Issue is blocked on a dependency that has since resolved	Click Override lock to release the dependency block manually.
Review loop caused by a bad issue or bad conventions	Fix .colony/conventions.md, then click Retry.
Issue should be abandoned	Click Cancel (equivalent to /colony:cancel).
DLQ entry cannot be auto-resolved	Investigate the error, then either retry or cancel the originating issue.

For a full list of slash commands available on individual issues and PRs, see the Operator role.