Drop 9 batch proposals that are superseded by the boocode-lift-analysis (boocontext-audit, conductor upgrades, self-healing/verify-gate skills): add-3tier-memory, import-llm-evaluator, import-pregel-engine, plugin-platform, conductor-evolution, code-intelligence-upgrade, dev-workflow, ui-overhaul, agent-reliability. Delete 11 stub archive files (49-66B each, 'Status: Shipped. Archived.' only) that provide zero documentation value over the existing CHANGELOG.md + git tags.
2.7 KiB
2.7 KiB
ADDED Requirements
Requirement: Token analytics page route
The web app SHALL expose a route at /analytics that renders the token analytics dashboard.
Scenario: Route resolves to Analytics page
- WHEN the user navigates to
/analytics - THEN the
Analyticscomponent renders without crashing - AND the sidebar nav button for "Token Analytics" shows as active
Requirement: Aggregate usage summary
The dashboard SHALL display an aggregate summary of token usage across all data sources.
Scenario: Summary cards show totals
- WHEN the analytics page loads
- THEN the page displays summary cards for: total input tokens, total output tokens, total cost (USD), and total sessions tracked
- AND each card shows a numeric value with a short label
Requirement: Per-session token breakdown
The dashboard SHALL display a list or table of recent sessions with per-session token and cost data.
Scenario: Session list renders
- WHEN
GET /api/coder/sessions/:id/agent-sessionsreturns data - THEN the page renders a table with columns: session name, agent/provider, input tokens, output tokens, cost, last active
Requirement: Per-tool cost breakdown
The dashboard SHALL display per-tool token consumption stats.
Scenario: Tool stats table renders
- WHEN the analytics page loads
- THEN it fetches and renders the per-tool cost stats from
/api/tools/cost_stats - AND displays: tool name, call count, avg prompt tokens, avg completion tokens, total avg tokens
Requirement: Context window utilization
The dashboard SHALL display context window utilization data across sessions where available.
Scenario: Context usage section renders
- WHEN session message data with
ctx_used/ctx_maxfields is available - THEN a section shows average context utilization percentage across sessions
Requirement: Token breakdown by category
The dashboard SHALL display token breakdown by category (system/user/assistant/tools/reasoning) where token_breakdown JSONB data exists.
Scenario: Category breakdown renders
- WHEN a session has
token_breakdowndata on its tasks - THEN the dashboard shows per-category token counts and a visual breakdown (e.g., stacked bar or pie segment)
Requirement: Loading and error states
The dashboard SHALL handle loading and error states gracefully.
Scenario: Loading state
- WHEN data is being fetched
- THEN a loading indicator is shown in each section
Scenario: Error state
- WHEN an API call fails
- THEN the affected section shows an error message and a retry button
- AND other sections continue to render independently