ticket-analysis

Step 0 — Load configuration

Read ${CLAUDE_PLUGIN_ROOT}/config.yaml. If it does not exist, tell the user to copy config.example.yaml to config.yaml and fill in their values. Stop here.

Extract:

• issue_tracker — determines which adapter to load
• design_tool — figma, penpot, or none
• docs_sources — array of documentation sources
• ticket_analysis.ui_keywords — keywords that indicate UI work
• Communication config for sending results

Step 1 — Load adapter

Read the adapter file at ${CLAUDE_PLUGIN_ROOT}/skills/ticket-analysis/adapters/{issue_tracker}.md. If the adapter does not exist, tell the user that {issue_tracker} is not yet supported and stop.

Core principles

1. Zero assumptions — Never infer what the ticket does not explicitly state. If something is unclear, mark it as a question.
2. Traceability — Every finding must reference its source (ticket field, design screen, doc page, code file).
3. Fixed structure — Always output the 8 sections defined in the output format, even if some are empty.

Phase 1 — Fetch and parse ticket

Use the adapter's fetch ticket command to retrieve {key}.

Extract:

• Title, description, acceptance criteria, status, type, assignee, labels
• URLs — all URLs in description and comments (design links, doc links, API specs)
• Data terms — entity names, field names, table names mentioned
• Referenced tickets — other ticket keys mentioned (pattern: [A-Z]+-\d+)
• Comments — all comments, in chronological order

Phase 1b — Classify ticket type

Based on title, description, and labels, classify the ticket as one of:

Type	Indicators
business-feature	User-facing value, UI work, business rules
technical-infra	Infrastructure, refactoring, performance, DevOps
bug-fix	Defect, error, unexpected behavior
data-integration	Data migration, API integration, ETL, sync

The classification determines which Phase 2 agents to launch.

Phase 2 — Parallel analysis

Launch up to 4 agents in parallel based on ticket type and available tools. Each agent is a sub-analysis.

Agent A: Design analysis

Launch when: {config.design_tool} is not "none" AND (ticket has design URLs OR title/description matches any {config.ticket_analysis.ui_keywords})

Use the analyze-design skill approach:

• Fetch the design using the design tool
• Extract data fields, states, actions
• Compare with ticket description

Agent B: Documentation analysis

Launch when: {config.docs_sources} is not empty

Use the analyze-docs skill approach:

• Search each documentation source for key terms from the ticket
• Extract relevant business rules, definitions, existing specs

Agent C: Data model analysis

Launch when: ticket mentions entities, fields, tables, or database operations

Use the analyze-data-model skill approach:

• Search codebase for mentioned entities
• Map fields, relationships, validations
• Identify gaps between ticket and existing model

Agent D: Technical feasibility

Launch when: ticket involves code changes

Use the analyze-feasibility skill approach:

• Check existing patterns in codebase
• Identify affected modules
• Assess implementation complexity

Phase 3 — Cross-reference

Combine results from all agents. For each data point (field, rule, state, action):

Build a cross-reference table:

Dato	Ticket	Diseño	Documentación	Código	Estado
{field}	{present/absent}	{present/absent}	{present/absent}	{present/absent}	{status}

Status values:

• ✅ Claro — consistent across all sources
• ❓ Dudoso — present in some sources but inconsistent or incomplete
• ❌ No existe — mentioned in ticket but not found in any other source

Phase 4 — Build questions

Group all questions by audience:

• Product / PO — business rules, scope, priority
• Design / UX — visual behavior, states, interactions
• Tech Lead — architecture, approach, data model
• QA — test scenarios, edge cases
• Data Team — data sources, transformations, validations

Each question must:

1. State what is unclear
2. Reference the source that triggered it
3. Suggest options when possible

Phase 5 — Define scope

Based on all gathered information, classify each requirement:

• SE ENTREGA — fully defined, design exists, data model supports it, technically feasible
• NO SE ENTREGA — explicitly out of scope or not mentioned in ticket
• BLOQUEADO — cannot implement without answers to blocking questions

Phase 6 — Next steps

Define concrete next steps:

1. Questions that must be answered before starting
2. Technical spikes needed
3. Design clarifications needed
4. Dependencies to unblock
5. Suggested subtask breakdown

Phase 7 — Output

Read the output format at references/output-format.md.

Build the report using the fixed 8-section structure defined there. Never skip sections — if a section has no findings, write "No se encontraron hallazgos en esta sección."

Phase 8 — Offer actions

1. Comment on ticket — Post a summary of findings and questions using the adapter's comment command
2. Update ticket — Add scope definition and questions to ticket description
3. Create subtasks — Break down into implementation subtasks using the adapter
4. Send to team — If communication tool is configured, send summary to team channel
5. Export — Display for copying