Skip to main content
← Back to list
01Issue
FeatureOpenSwamp ClubTeam
AssigneesNone

Relationships

#1042 Collective leaderboard: activity metrics aggregated by identity group

Opened by webframp · 7/8/2026

Problem

Swamp already captures per-principal activity through audit logs, method history, and versioned data with owner attribution. It also has a group and collective identity model (access group, access can-i --collectives, IdP group mappings). But there is no way to aggregate activity metrics by collective and surface ranked summaries. Operators managing a shared swamp server have no visibility into which collectives are most active, what their workload profiles look like, or how usage distributes across teams.

Proposed solution

A leaderboard feature that aggregates activity metrics by collective (IdP group or local group) and publishes the results as queryable typed data.

Metrics to aggregate per collective

  • Method runs — total runs, unique models touched, success/failure counts
  • Data volume — data entries produced, total size, data types distribution
  • Workflow activity — workflow runs, steps executed, completion rates
  • Model management — models created, methods invoked by type
  • Active principals — unique principals within the collective over the reporting window

Data model

The leaderboard output should be typed swamp data, versioned and queryable via CEL like any other data. This means downstream models and workflows can consume leaderboard metrics directly — for example, a workflow could gate on "collective X has run fewer than N deployments this week" as a CEL condition.

Reporting windows

Support configurable time windows (last 24h, 7d, 30d) so the leaderboard captures both recent activity and longer-term trends.

Access integration

Respect the existing grant model. Principals should see metrics for their own collectives by default. Server admins see all collectives. The --collectives flag on access can-i already models this boundary.

Alternatives considered

  • Report extension only: A report extension could query audit and method history to produce similar output. This works but requires each operator to build their own aggregation logic. A native feature standardizes the metrics and makes them available without extension authorship.
  • External dashboard: Exporting audit logs to an external system (Grafana, Honeycomb) for aggregation. This breaks the "data is queryable from within swamp" principle and requires infrastructure outside the swamp ecosystem.

Why this matters

Collectives are already a first-class identity concept in swamp's access model. Surfacing activity metrics by collective closes the loop: groups are not just authorization boundaries but observable units of work. This turns the access model into an activity model — who can do what becomes who is doing what, queryable through the same CEL primitives that wire everything else together.

02Bog Flow
◉OPEN○TRIAGED○IN PROGRESS○SHIPPED

Open

7/8/2026, 9:00:51 PM

No activity in this phase yet.

03Sludge Pulse

Sign in to post a ripple.