@riotprompt/riotplan 1.0.4 → 1.0.5

package/dist/index.d.ts

@@ -556,6 +556,11 @@ export declare interface ElaborationRecord {
     content: string;
 }
 
+/**
+ * Method used to gather evidence
+ */
+declare type EvidenceGatheringMethod = "manual" | "model-assisted";
+
 /**
  * Evidence record for inception phase materials
  *
@@ -579,6 +584,10 @@ export declare interface EvidenceRecord {
     summary?: string;
     /** Tags for categorization */
     tags?: string[];
+    /** How evidence was gathered */
+    gatheringMethod?: EvidenceGatheringMethod;
+    /** Relevance score (0-1) from model if model-assisted */
+    relevanceScore?: number;
 }
 
 /**

package/dist/mcp/prompts/create_plan.md

@@ -0,0 +1,127 @@
+# Create Plan Workflow
+
+You are helping the user create a new plan with AI-generated, actionable steps for a complex task or feature.
+
+## Your Task
+
+Follow this workflow to create a comprehensive plan using the riotplan MCP tools available to you.
+
+## Step 1: Review Provided Information
+
+Check what information has already been provided as prompt arguments:
+- **code**: ${code}
+- **description**: ${description}
+- **directory**: ${directory}
+- **steps**: ${steps}
+
+## Step 2: Gather Missing Information
+
+For any information marked as "[code]", "[description]", "[directory]", or "[steps]", ask the user to provide it:
+
+1. **Plan Code** (if missing) - Ask for a short identifier (e.g., "auth-system", "dark-mode", "refactor-db")
+2. **Plan Description** (if missing) - Ask for a clear, detailed description of what they want to accomplish
+3. **Target Directory** (if missing) - Ask where to create the plan. Suggest using a `plans/` directory if one exists, or offer to create one. If they don't specify, default to current directory.
+4. **Number of Steps** (if missing) - Ask how many steps to generate, or let the AI determine automatically
+5. **AI Provider** (optional) - Which provider to use (anthropic, openai, gemini). Default to anthropic if not specified.
+
+## Step 3: Create the Plan Using MCP Tools
+
+IMPORTANT: Use the `riotplan_create` MCP tool to create the plan. DO NOT shell out to CLI commands.
+
+Once you have all the required information (code, description, directory), call the `riotplan_create` tool:
+
+```
+{
+  "code": "the-plan-code",
+  "name": "Human Readable Name (optional)",
+  "description": "Detailed description of what to accomplish",
+  "directory": "path/to/parent/directory",
+  "steps": 8,
+  "provider": "anthropic",
+  "direct": false
+}
+```
+
+Parameters:
+- `code` (required) - Plan identifier
+- `description` (required) - What to accomplish
+- `name` (optional) - Human-readable name
+- `directory` (optional) - Parent directory for the plan
+- `steps` (optional) - Number of steps to generate
+- `direct` (optional) - Set to true to skip analysis phase
+- `provider` (optional) - AI provider (anthropic, openai, gemini)
+- `model` (optional) - Specific model to use
+- `noAi` (optional) - Use templates only, no AI generation
+
+## Step 4: Review the Generated Plan
+
+After creation, use the `riotplan_status` tool to check the plan:
+
+```
+{
+  "path": "./path/to/plan",
+  "verbose": true
+}
+```
+
+Inform the user about:
+- The plan location
+- Number of steps generated
+- The plan structure (SUMMARY.md, EXECUTION_PLAN.md, step files)
+
+## Step 5: Validate the Plan
+
+Use the `riotplan_validate` tool to ensure the plan structure is correct:
+
+```
+{
+  "path": "./path/to/plan"
+}
+```
+
+If validation fails, report the issues to the user.
+
+## Step 6: Prepare for Execution
+
+Inform the user that they can:
+- Review the generated plan files
+- Start the first step using `riotplan_step_start`
+- Track progress using `riotplan_status`
+- Add or modify steps as needed
+
+## Important Guidelines
+
+- **Always use MCP tools** - Never shell out to CLI commands like `riotplan create`
+- **Ask about directory** - Don't assume where the plan should be created
+- **Be specific** - Encourage detailed descriptions for better plan generation
+- **Right-size steps** - Suggest 5-10 steps for most tasks
+- **Use analysis** - Don't set `direct: true` unless the user specifically requests it
+- **Validate** - Always validate the plan after creation
+
+## Example Interaction
+
+User: "I want to build a dog behavior tracking system"
+
+You should:
+1. Ask clarifying questions:
+   - "What specific features should the system include?"
+   - "Where would you like me to create the plan? I can create it in a `plans/` directory if you'd like."
+   - "How many steps would you like? I recommend 8-10 for a system like this."
+
+2. Once you have the information, call `riotplan_create`:
+   ```
+   {
+     "code": "dog-tracker",
+     "name": "Dog Behavior and Training Tracking System",
+     "description": "Build a comprehensive dog behavior and training tracking system. The system should allow users to log daily behaviors, track training sessions, monitor progress over time, set training goals, and generate insights about patterns. Include features for multiple dogs, behavior categorization, training exercises library, progress visualization, and reporting capabilities.",
+     "directory": "./plans",
+     "steps": 10,
+     "provider": "anthropic"
+   }
+   ```
+
+3. Review the result with `riotplan_status`
+
+4. Validate with `riotplan_validate`
+
+5. Inform the user the plan is ready and they can begin execution