Triage

Triage is an AI-powered analysis engine that evaluates incoming Jira issues before they enter the Clears AI pipeline. It scores each issue on risk, complexity, and agent confidence so your team can prioritize what to import and decide which stories are good candidates for AI implementation.

How Triage Works

When triage is enabled, Clears AI automatically analyzes issues from your connected Jira projects. For each issue, the AI evaluates:

• Risk Level - How risky is this change? Considers blast radius, data sensitivity, and architectural impact.
• Complexity - How complex is the implementation? Considers cross-service dependencies, unknowns, and scope.
• Agent Confidence - How confident is the AI that it can implement this story autonomously? Expressed as a percentage (0–100%).

Confidence Bands

Agent confidence scores map to four bands:

Band	Score Range	Meaning
High	90–100%	The AI is highly confident it can implement this story end-to-end
Medium	70–89%	The AI can likely handle this with some guidance or Q&A
Low	Below 70%	The AI may struggle, consider manual implementation or story decomposition
Not triaged	—	The issue has not yet been analyzed

Configuring Triage Settings

Navigate to Settings > Triage to configure the triage engine.

Triage Status

Toggle triage on or off for your workspace. When enabled, issues matching your filters are automatically analyzed.

Run Behavior

Select a workflow to trigger when triage completes. Use the Run Behavior dropdown to choose which workflow (if any) should execute after an issue is triaged. Leave empty to analyze without triggering further automation.

Filters

Narrow which Jira issues are triaged using these filters:

Filter	Description
Project	Jira project keys to include (e.g., PROJ, BACKEND)
Reporter	Only triage issues from specific reporters
Assignee	Only triage issues assigned to specific users
Issue Type	Restrict to specific types: Story, Task, or Bug

Viewing Triage Results

Triage results appear in the Import from Jira view on the Board. Each issue row displays:

Column	Description
Relevant Repos	Repositories the AI identified as related to the issue
Risk Level	Low, Medium, or High
Complexity	Low, Medium, or High
Agent Confidence	A percentage with a progress bar indicating the confidence band

Expanding Issue Details

Click on any triaged issue to expand its detail panel. The expanded view shows:

• Problem / Need - A concise AI-generated summary of the issue's core requirement.
• Risk Reasoning - Why the AI assigned the given risk level.
• Complexity Reasoning - Why the AI assigned the given complexity level.
• Confidence Reasoning - What factors influenced the agent confidence score.
• Relevant Repos - Each repository listed with its individual confidence score (0.0–1.0).

Triage API

Triage settings and results are also accessible via the REST API:

Endpoint	Method	Description
/triage/settings	GET	Retrieve current triage configuration
/triage/settings	PUT	Update triage configuration
/triage/results/{story_key}	GET	Retrieve triage results for a specific issue

Settings Object

{
  "enabled": true,
  "project_keys": ["PROJ", "BACKEND"],
  "reporter_ids": [],
  "assignee_ids": [],
  "issue_types": ["story", "task", "bug"],
  "run_workflow_id": "wf_abc123"
}

Result Object

{
  "risk_level": "medium",
  "complexity": "high",
  "agent_confidence": 82,
  "problem_need": "Add rate limiting to the public API gateway",
  "risk_reasoning": "Touches shared infrastructure used by multiple services",
  "complexity_reasoning": "Requires changes across API gateway, auth service, and config",
  "confidence_reasoning": "Well-documented patterns exist but cross-service coordination adds uncertainty",
  "relevant_repos": [
    "https://github.com/org/api-gateway",
    "https://github.com/org/auth-service"
  ],
  "relevant_repos_confidence": {
    "https://github.com/org/api-gateway": 0.95,
    "https://github.com/org/auth-service": 0.72
  }
}