Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.ironbee.ai/llms.txt

Use this file to discover all available pages before exploring further.

Findings are observations that IronBee’s analysis identified — a browser console error that kept appearing, a failing form submission, a file with a high fix-cycle rate, or a pattern of retries. Each finding is grounded in actual session data, not generic advice.

Finding areas

Findings are grouped by area depending on which analysis type produced them:
AreaQuestion answeredAnalysis type
qualityIs the agent testing thoroughly enough?Quality
efficiencyWhere is time or compute being wasted?Quality, Cost
patternsWhich issues and files recur across sessions?Quality
behaviorWhat working habits need adjustment?Quality, Session Insights
costWhere is spend going and what should change?Cost

Severity levels

SeverityMeaning
noticeDescriptive observation — informational only, no action required
concernWorth watching — may generate a recommendation
criticalSerious degradation — should generate a recommendation
Only concern and critical findings may have recommendations attached. notice findings are purely descriptive.

Findings list

The Findings tab shows all findings for the current analysis, grouped by area and sorted by severity. Each finding includes:
  • Title — a brief description of the observation
  • Severity — critical / concern / notice
  • Area — quality / efficiency / patterns / behavior / cost
  • Body — full markdown description with supporting evidence, metrics, and sometimes visualizations
  • Linked recommendation (if any) — the action derived from this finding

Visualizations

Some findings include inline diagrams (pie charts, bar charts, Gantt, journey diagrams) rendered from the finding body. Visualizations are decorative — every fact they show is also stated in prose.

Session insight findings

Session insight findings are rich multi-paragraph reports rather than single observations. They cover:
  • Session highlight — a standout session with notable characteristics
  • At-a-glance — four summary sub-findings (sessions, hours, cost, cache)
  • Project areas — patterns, friction points, interaction style
  • What works — validated effective behaviors
  • Features to try — underused capabilities relevant to the project

Project-level findings

The project Analysis tab always shows the latest analysis run. Previous runs are retained as history — you can navigate back to compare findings over time. Files that appear in findings most often are strong candidates for extra attention — better tests, clearer logic, or structural refactoring.