Skip to content

tools.md

Latest commit

 

History

History
132 lines (92 loc) · 5.69 KB

tools.md

File metadata and controls

132 lines (92 loc) · 5.69 KB
Raw

MCP Tool Catalog

This document describes the MCP servers used by the sdlc-workflow plugin skills. All tool references use the generic naming convention — see your project's CLAUDE.md for concrete instance names.

Jira MCP (Atlassian)

Purpose: Task management, issue tracking, and workflow state.

Tool prefix: mcp__atlassian__

Access method: MCP-first with REST API v3 fallback

Key Operations

Operation Tool Used By
Fetch issue details mcp__atlassian__getJiraIssue plan-feature, implement-task
Create issue mcp__atlassian__createJiraIssue plan-feature
Edit issue fields mcp__atlassian__editJiraIssue implement-task
Add comment mcp__atlassian__addCommentToJiraIssue plan-feature, implement-task
Transition issue mcp__atlassian__transitionJiraIssue implement-task
Get transitions mcp__atlassian__getTransitionsForJiraIssue implement-task
Create issue link mcp__atlassian__createIssueLink plan-feature
Get current user mcp__atlassian__atlassianUserInfo implement-task
Search issues (JQL) mcp__atlassian__searchJiraIssuesUsingJql plan-feature

REST API Fallback

When Atlassian MCP is unavailable (e.g., due to organizational policies restricting localhost access), skills automatically fall back to JIRA REST API v3.

How it works:

  1. Skills always try MCP first (preferred method)
  2. If MCP fails, user is always prompted to choose: Use REST API, Skip JIRA, or Retry MCP
  3. If user chooses REST API:
    • Skills check CLAUDE.md for existing REST API credentials
    • If credentials exist: Use them (with user confirmation)
    • If credentials don't exist: Collect from user, validate, and optionally store
  4. All operations use scripts/jira-client.py (Python stdlib only, no external dependencies)

Credential storage: Credentials are stored in CLAUDE.md under ## Jira Configuration → ### REST API Credentials (MCP Fallback). Users can choose:

  • Store all in CLAUDE.md (convenient, less secure)
  • Store URL/email only, use $JIRA_API_TOKEN env var (recommended, more secure)
  • Don't store (ask each time)

Documentation:

  • Setup guide: plugins/sdlc-workflow/shared/jira-api-token-guide.md
  • Implementation guide: plugins/sdlc-workflow/shared/jira-rest-fallback.md
  • Access strategy: plugins/sdlc-workflow/shared/jira-access-strategy.md

ADF Note

Several Jira operations require Atlassian Document Format (ADF) rather than plain text. This applies to:

  • Issue descriptions (when using contentFormat: "adf")
  • Comments (when using contentFormat: "adf")
  • Custom fields (e.g., Git Pull Request custom field)

The REST API fallback automatically converts markdown to ADF when needed.

Refer to your project's CLAUDE.md for field IDs and formatting details specific to your Jira instance.

Figma MCP

Purpose: Design inspection during the planning phase.

Tool prefix: mcp__figma__

Key Operations

Operation Tool Used By
Get file structure mcp__figma__get_file plan-feature
Get specific nodes mcp__figma__get_nodes plan-feature

Usage

Figma is used exclusively during plan-feature to extract:

  • UI structure and component hierarchy
  • Components involved in the feature
  • Possible backend interactions implied by the design

Figma URLs can be provided as a skill argument or discovered from the Jira issue description.

Serena LSP MCP

Purpose: Code intelligence — symbol search, impact analysis, and symbolic editing.

Tool prefix: mcp__<instance>__ where <instance> is the Serena instance name from the Repository Registry in your project's CLAUDE.md.

Each target repository has a dedicated Serena MCP server instance. Look up the correct instance name in the Repository Registry table of your project's CLAUDE.md.

Key Operations

Operation Tool Used By
File/module structure overview mcp__<instance>__get_symbols_overview plan-feature, implement-task
Find symbols by name mcp__<instance>__find_symbol plan-feature, implement-task
Find callers/references mcp__<instance>__find_referencing_symbols plan-feature, implement-task
Search for patterns mcp__<instance>__search_for_pattern plan-feature, implement-task
Replace symbol body mcp__<instance>__replace_symbol_body implement-task
Insert code after symbol mcp__<instance>__insert_after_symbol implement-task
Insert code before symbol mcp__<instance>__insert_before_symbol implement-task
Rename symbol mcp__<instance>__rename_symbol implement-task
Get diagnostics mcp__<instance>__get_diagnostics setup

Workflow

During planning (read-only):

  1. get_symbols_overview — understand module structure without reading entire files
  2. find_symbol — locate types, functions, endpoints by name
  3. find_referencing_symbols — discover callers and integration points
  4. search_for_pattern — find string literals, config keys, route definitions

During implementation (read + write):

  1. Use the same read operations to understand code before modifying
  2. replace_symbol_body — rewrite functions, methods, structs, or components
  3. insert_after_symbol / insert_before_symbol — add new code relative to existing symbols
  4. rename_symbol — rename and automatically update all references

Fallback

If no Serena instance is available for a repository, skills fall back to Read, Grep, and Glob tools for code inspection, and Edit/Write tools for modification.

Limitations

Check the Code Intelligence > Limitations section in your project's CLAUDE.md for per-instance limitations (e.g., language server features that are not supported).