gobby 0.2.5__py3-none-any.whl

gobby/prompts/defaults/features/import_mcp_github.md

@@ -0,0 +1,17 @@
+---
+name: features-import-mcp-github
+description: User prompt for GitHub-based MCP server import
+version: "1.0"
+variables:
+  github_url:
+    type: str
+    required: true
+    description: The GitHub repository URL
+---
+Fetch the README from this GitHub repository and extract MCP server configuration:
+
+{{ github_url }}
+
+If the URL doesn't point directly to a README, try to find and fetch the README.md file.
+
+After reading the documentation, extract the MCP server configuration as a JSON object.

gobby/prompts/defaults/features/import_mcp_search.md

@@ -0,0 +1,16 @@
+---
+name: features-import-mcp-search
+description: User prompt for search-based MCP server import
+version: "1.0"
+variables:
+  search_query:
+    type: str
+    required: true
+    description: The search query for finding the MCP server
+---
+Search for MCP server: {{ search_query }}
+
+Find the official documentation or GitHub repository for this MCP server.
+Then fetch and read the README or installation docs.
+
+After reading the documentation, extract the MCP server configuration as a JSON object.

gobby/prompts/defaults/features/recommend_tools.md

@@ -0,0 +1,32 @@
+---
+name: features-recommend-tools
+description: System prompt for MCP server tool recommendations
+version: "1.0"
+---
+You are a tool recommendation assistant for Claude Code with access to MCP servers.
+
+CRITICAL PRIORITIZATION RULES:
+1. Analyze the task type (code navigation, docs lookup, database query, planning, data processing, etc.)
+2. Check available MCP server DESCRIPTIONS for capability matches
+3. If ANY MCP server's description matches the task type -> recommend those tools FIRST
+4. Only recommend built-in Claude Code tools (Grep, Read, Bash, WebSearch) if NO suitable MCP server exists
+
+TASK TYPE MATCHING GUIDELINES:
+- Task needs library/framework documentation -> Look for MCP servers describing "documentation", "library docs", "API reference"
+- Task needs code navigation/architecture understanding -> Look for MCP servers describing "code analysis", "symbols", "semantic search"
+- Task needs database operations -> Look for MCP servers describing "database", "PostgreSQL", "SQL"
+- Task needs complex reasoning/planning -> Look for MCP servers describing "problem-solving", "thinking", "reasoning"
+- Task needs data processing/large datasets -> Look for MCP servers describing "code execution", "data processing", "token optimization"
+
+ANTI-PATTERNS (What NOT to recommend):
+- Don't recommend WebSearch when an MCP server provides library/framework documentation
+- Don't recommend Grep/Read for code architecture questions when an MCP server does semantic code analysis
+- Don't recommend Bash for database queries when an MCP server provides database tools
+- Don't recommend direct implementation when an MCP server provides structured reasoning
+
+OUTPUT FORMAT:
+Be concise and specific. Recommend 1-3 tools maximum with:
+1. Which MCP server and tools to use (if applicable)
+2. Brief rationale based on server description matching task type
+3. Suggested workflow (e.g., "First call X, then use result with Y")
+4. Only mention built-in tools if no MCP server is suitable

gobby/prompts/defaults/features/recommend_tools_hybrid.md

@@ -0,0 +1,35 @@
+---
+name: features-recommend-tools-hybrid
+description: Prompt for hybrid mode re-ranking of semantic search results
+version: "1.0"
+variables:
+  task_description:
+    type: str
+    required: true
+    description: The task description to match tools for
+  candidate_list:
+    type: str
+    required: true
+    description: Formatted list of candidate tools from semantic search
+  top_k:
+    type: int
+    default: 5
+    description: Number of top recommendations to return
+---
+You are an expert at selecting tools for tasks.
+Task: {{ task_description }}
+
+Candidate tools (ranked by semantic similarity):
+{{ candidate_list }}
+
+Re-rank these tools by relevance to the task and provide reasoning.
+Return the top {{ top_k }} most relevant as JSON:
+{
+  "recommendations": [
+    {
+      "server": "server_name",
+      "tool": "tool_name",
+      "reason": "Why this tool is the best choice"
+    }
+  ]
+}

gobby/prompts/defaults/features/recommend_tools_llm.md

@@ -0,0 +1,30 @@
+---
+name: features-recommend-tools-llm
+description: Prompt for pure LLM-based tool recommendations
+version: "1.0"
+variables:
+  task_description:
+    type: str
+    required: true
+    description: The task description to match tools for
+  available_servers:
+    type: str
+    required: true
+    description: Formatted list of available MCP servers with descriptions
+---
+You are an expert at selecting the right tools for a given task.
+Task: {{ task_description }}
+
+Available Servers: {{ available_servers }}
+
+Please recommend which tools from these servers would be most useful for this task.
+Return a JSON object with this structure:
+{
+  "recommendations": [
+    {
+      "server": "server_name",
+      "tool": "tool_name",
+      "reason": "Why this tool is useful"
+    }
+  ]
+}

gobby/prompts/defaults/features/server_description.md

@@ -0,0 +1,20 @@
+---
+name: features-server-description
+description: Prompt for generating MCP server descriptions from tools
+version: "1.0"
+variables:
+  server_name:
+    type: str
+    required: true
+    description: The MCP server name
+  tools_list:
+    type: str
+    required: true
+    description: Formatted list of tools on the server
+---
+Write a single concise sentence describing what the '{{ server_name }}' MCP server does based on its tools.
+
+Tools:
+{{ tools_list }}
+
+Description (1 sentence, try to keep under 100 characters):

gobby/prompts/defaults/features/server_description_system.md

@@ -0,0 +1,6 @@
+---
+name: features-server-description-system
+description: System prompt for server description generation
+version: "1.0"
+---
+You write concise technical descriptions.

gobby/prompts/defaults/features/task_description.md

@@ -0,0 +1,31 @@
+---
+name: features-task-description
+description: Prompt for generating task descriptions from spec sections
+version: "1.0"
+variables:
+  task_title:
+    type: str
+    required: true
+    description: The task title
+  section_title:
+    type: str
+    default: ""
+    description: The spec section title
+  section_content:
+    type: str
+    default: ""
+    description: The spec section content
+  existing_context:
+    type: str
+    default: ""
+    description: Any existing context about the task
+---
+Generate a concise task description for this task from a spec document.
+
+Task title: {{ task_title }}
+Section: {{ section_title }}
+Section content: {{ section_content }}
+Existing context: {{ existing_context }}
+
+Write a 1-2 sentence description focusing on the goal and deliverable.
+Do not add quotes, extra formatting, or implementation details.

gobby/prompts/defaults/features/task_description_system.md

@@ -0,0 +1,6 @@
+---
+name: features-task-description-system
+description: System prompt for task description generation
+version: "1.0"
+---
+You are a technical writer creating concise task descriptions for developers.

gobby/prompts/defaults/features/tool_summary.md

@@ -0,0 +1,17 @@
+---
+name: features-tool-summary
+description: Prompt for summarizing MCP tool descriptions
+version: "1.0"
+variables:
+  description:
+    type: str
+    required: true
+    description: The tool description to summarize
+---
+Summarize this MCP tool description in 180 characters or less.
+Keep it to three sentences or less. Be concise and preserve the key functionality.
+Do not add quotes, extra formatting, or code examples.
+
+Description: {{ description }}
+
+Summary:

gobby/prompts/defaults/features/tool_summary_system.md

@@ -0,0 +1,6 @@
+---
+name: features-tool-summary-system
+description: System prompt for tool description summarization
+version: "1.0"
+---
+You are a technical summarizer. Create concise tool descriptions.

gobby/prompts/defaults/research/step.md

@@ -0,0 +1,58 @@
+---
+name: research-step
+description: ReAct loop prompt for codebase research
+version: "1.0"
+variables:
+  task_title:
+    type: str
+    required: true
+    description: Task title being researched
+  task_description:
+    type: str
+    default: ""
+    description: Task description
+  found_files:
+    type: list
+    default: []
+    description: List of files found so far
+  snippets_keys:
+    type: list
+    default: []
+    description: List of snippet keys/paths found
+  history:
+    type: str
+    default: ""
+    description: Formatted history of recent agent turns
+  step:
+    type: int
+    required: true
+    description: Current step number (1-indexed)
+  max_steps:
+    type: int
+    default: 10
+    description: Maximum number of steps
+  search_tool_section:
+    type: str
+    default: ""
+    description: Optional web search tool description if available
+---
+Task: {{ task_title }}
+Description: {{ task_description }}
+
+You are researching this task to identify relevant files and implementation details.
+You have access to the following tools:
+
+1. glob(pattern): Find files matching a pattern (e.g. "src/**/*.py")
+2. grep(pattern, path): Search for text in files (e.g. "def login", "src/")
+3. read_file(path): Read the content of a file
+4. done(reason): Finish research
+{{ search_tool_section }}
+
+Current Context:
+Found Files: {{ found_files }}
+Snippets: {{ snippets_keys }}
+
+History:
+{{ history }}
+
+Step {{ step }}/{{ max_steps }}. What is your next move? Respond with THOUGHT followed by ACTION.

gobby/prompts/defaults/validation/criteria.md

@@ -0,0 +1,47 @@
+---
+name: validation-criteria
+description: Prompt for generating validation criteria from task description
+version: "1.0"
+variables:
+  title:
+    type: str
+    required: true
+    description: Task title
+  description:
+    type: str
+    default: "(no description)"
+    description: Task description
+---
+Generate validation criteria for this task.
+
+Task: {{ title }}
+Description: {{ description }}
+
+CRITICAL RULES - You MUST follow these:
+1. **Only stated requirements** - Include ONLY requirements explicitly written in the title or description
+2. **No invented values** - Do NOT invent specific numbers, timeouts, thresholds, or limits unless they appear in the task
+3. **No invented edge cases** - Do NOT add edge cases, error scenarios, or boundary conditions beyond what's described
+4. **Proportional detail** - Vague tasks get vague criteria; detailed tasks get detailed criteria
+5. **When in doubt, leave it out** - If something isn't mentioned, don't include it
+
+For vague requirements like "fix X" or "add Y", use criteria like:
+- "X no longer produces the reported error/warning"
+- "Y functionality works as expected"
+- "Existing tests continue to pass"
+- "No regressions introduced"
+
+DO NOT generate criteria like:
+- "timeout defaults to 30 seconds" (unless 30 seconds is in the task description)
+- "handles edge case Z" (unless Z is mentioned in the task)
+- "logs with format X" (unless that format is specified)
+
+Format as markdown checkboxes:
+## Deliverable
+- [ ] What the task explicitly asks for
+
+## Functional Requirements
+- [ ] Only requirements stated in the description
+
+## Verification
+- [ ] Tests pass (if applicable)
+- [ ] No regressions

gobby/prompts/defaults/validation/validate.md

@@ -0,0 +1,38 @@
+---
+name: validation-validate
+description: Base prompt for validating task completion against criteria
+version: "1.0"
+variables:
+  title:
+    type: str
+    required: true
+    description: Task title
+  category_section:
+    type: str
+    default: ""
+    description: Optional category/test strategy section
+  criteria_text:
+    type: str
+    required: true
+    description: Validation criteria or task description
+  changes_section:
+    type: str
+    required: true
+    description: Summary of changes made (files, diffs, etc.)
+  file_context:
+    type: str
+    default: ""
+    description: Optional file content context
+---
+Validate if the following changes satisfy the requirements.
+
+Task: {{ title }}
+{{ category_section }}{{ criteria_text }}
+
+{{ changes_section }}
+IMPORTANT: Return ONLY a JSON object, nothing else. No explanation, no preamble.
+Format: {"status": "valid", "feedback": "..."} or {"status": "invalid", "feedback": "..."}
+{% if file_context %}
+File Context:
+{{ file_context }}
+{% endif %}