feat(workflow): enhance agent input/output definitions and validation for multi-agent workflows

Author: moazbuilds

prompts/templates/ali/chained/step-04-prompts.md

@@ -10,6 +10,7 @@ description: "Create prompt files and register placeholders"
 Create the actual prompt files for all agents:
 - Create folder structure
 - Generate prompt files for each agent
+- Define input and output for each agent
 - Create shared files
 - Register NEW placeholders in config
 
@@ -260,6 +261,25 @@ Remove previous agent output from list.
 
 **For all agents, after context decided:**
 
+"**What output does '\{agent.name\}' produce?**
+
+Since agents are isolated and can't see each other's work, we need to define what this agent outputs so the next agent can receive it.
+
+*[If agent produces artifacts/plans/code:]*
+I recommend: `.codemachine/artifacts/\{agent.id\}-output.md`
+
+*[If Q&A/interactive agent that only collects info:]*
+This agent collects user input - it may not need a file output if the next agent can access conversation context.
+
+**Output filename** (or 'none' for Q&A agents):"
+
+Wait. Store as `agent.outputFile`.
+
+*[If output file specified, confirm placeholder:]*
+"This will be registered as placeholder `\{agent_id\}_output` for the next agent to receive."
+
+---
+
 "**Should this agent's prompt be split into smaller files?**
 
 Based on '\{agent.name\}', I \{recommend/don't recommend\} splitting because \{reasoning\}.
@@ -306,18 +326,28 @@ Moving to '\{next_agent.name\}'..."
 
 "**Context Summary:**"
 
-| Agent | Input From | Outputs To | Placeholders |
-|-------|------------|------------|--------------|
-| \{agent1\} | - | \{agent2\} | \{list\} |
-| \{agent2\} | \{agent1\} | \{agent3\} | \{list\} |
+| Agent | Input Type | Input Source | Output File | Output Placeholder |
+|-------|------------|--------------|-------------|-------------------|
+| \{agent1\} | \{type\} | \{source or '-'\} | \{file\} | \{placeholder\} |
+| \{agent2\} | placeholder | \{agent1_output\} | \{file\} | \{placeholder\} |
 
 *[Show data flow visually:]*
 
 "**Data Flow:**
 ```
-\{agent1\} → \{agent2\} → \{agent3\}
+\{agent1\} → writes \{output-file\} → registered as \{placeholder\}
+     ↓
+\{agent2\} → receives \{placeholder\} → writes \{output-file\} → registered as \{placeholder\}
+     ↓
+\{agent3\} → receives \{placeholder\} → ...
 ```"
 
+*[If any agent has no input defined (except first agent):]*
+"**Note:** The following agents have no input defined - they won't receive context from previous agents:
+- \{list\}
+
+Is this intentional? (Q&A agents may not need input from previous agents)"
+
 "**Shared files to create:**"
 - \{deduplicated list\}
 
@@ -1436,7 +1466,10 @@ For each shared:
       <persona path="prompts/templates/\{workflow_name\}/\{agent-id\}/persona.md" created="true" />
       <prompt path="prompts/templates/\{workflow_name\}/\{agent-id\}/\{prompt.md or workflow.md\}" created="true">
         <goal>\{agent goal\}</goal>
-        <output>\{expected output\}</output>
+        <input-type>\{placeholder|codebase-read|user-qa|specification\}</input-type>
+        <input-source>\{placeholder name or description\}</input-source>
+        <output-file>\{output filename or 'none'\}</output-file>
+        <output-placeholder>\{placeholder name for next agent\}</output-placeholder>
         <has-sub-agent-coordination>\{true/false\}</has-sub-agent-coordination>
       </prompt>
       <!-- For each chained step (if multi-step) -->
@@ -1517,6 +1550,11 @@ For each shared:
     <controller>\{Yes/No\}</controller>
     <shared-files>\{count\}</shared-files>
     <total-files>\{count\}</total-files>
+    <agent-chain>
+      <agents-with-input>\{count\}</agents-with-input>
+      <agents-with-output>\{count\}</agents-with-output>
+      <placeholder-chains>\{count\}</placeholder-chains>
+    </agent-chain>
   </summary>
 </step-04>
 ```

prompts/templates/ali/chained/step-05-workflow-generation.md

@@ -506,6 +506,27 @@ After appending:
 - Verify all used placeholders are registered
 - Result: "✓ All placeholders registered" or "⚠️ Unregistered: \{name\}"
 
+**Check 5: Agent Input/Output Chain** (for multi-agent workflows)
+- For workflows with 2+ agents, verify each agent has defined input/output
+- Verify agent outputs are registered as placeholders for subsequent agents
+- Verify each non-first agent has input defined (placeholder from previous agent, codebase read, or user Q&A)
+- Exception: Q&A/interactive agents that only collect user input don't need file output
+- Result: "✓ Agent chain properly connected" or "⚠️ Agent '\{id\}' has no input defined - it won't receive context from previous agents"
+
+*[If Check 5 fails:]*
+"**Agent chain incomplete.**
+
+The following agents have no defined input - they won't know what previous agents produced:
+\{list agents without input\}
+
+Options:
+1. Go back to Step 04 and define input/output for these agents
+2. Confirm these are standalone agents that don't need previous context
+
+How would you like to proceed?"
+
+Wait for response. Do not generate workflow until resolved.
+
 **If any issues found:**
 "**Issues Found:**"
 List each issue.
@@ -588,6 +609,7 @@ codemachine workflow \{workflow_name\}
     <files-exist>\{true|false\}</files-exist>
     <workflow-integrity>\{true|false\}</workflow-integrity>
     <placeholders-registered>\{true|false\}</placeholders-registered>
+    <agent-chain-connected>\{true|false\}</agent-chain-connected>
     <issues-fixed count="\{count\}">
       <!-- If any issues were fixed -->
       <issue type="\{type\}" path="\{path\}" action="\{action taken\}" />
@@ -639,7 +661,8 @@ Run: `codemachine workflow \{workflow_name\}`"
 - Missing files recreated if any
 - Workflow.js file created
 - All config files updated (main.agents.js, sub.agents.js, modules.js, placeholders.js)
-- Final validation passed (IDs, files, integrity, placeholders)
+- Final validation passed (IDs, files, integrity, placeholders, agent chain)
+- Agent input/output chain validated for multi-agent workflows
 - User educated about keyboard shortcuts
 - Run command provided
 - **Step-05 XML appended to plan file**
@@ -651,6 +674,7 @@ Run: `codemachine workflow \{workflow_name\}`"
 - Workflow.js not created
 - Config files not updated
 - Validation failures not addressed
+- Proceeding with multi-agent workflow where agents have no input defined (they won't receive context)
 - Not showing how to run the workflow
 - Skipping keyboard shortcuts education
 - **Not appending to plan file**

prompts/templates/ali/workflow.md

@@ -222,6 +222,39 @@ When user chooses to brainstorm in Step 01, select techniques from this list bas
 
 {ali_brain_methods}
 
+## Agent Isolation & Placeholders
+
+Each agent runs in isolation - it has no knowledge of what agents came before or after it, or what they produced. Placeholders are the mechanism for passing context between agents.
+
+Without placeholders, a multi-agent workflow is just disconnected agents that can't build on each other's work.
+
+### Input & Output
+
+Every agent should define:
+- **Input**: How the agent receives context
+- **Output**: What the agent produces (with exact filename so it can become a placeholder)
+
+| Input Type | Use Case |
+|------------|----------|
+| Placeholder | Receiving output from previous agent (`{planner_output}`) |
+| Codebase Read | Agent collects context by reading files |
+| User Q&A | Interactive agent - input comes from user answers |
+| Specification | First agent receiving initial spec (`{specification}`) |
+
+| Output Type | Description |
+|-------------|-------------|
+| Named Output | Specify exact filename: `{agent-id}-output.md` - becomes placeholder for next agent |
+| No Output | Only for Q&A/interactive agents that collect info but don't produce artifacts |
+
+### How It Works
+
+```
+Agent A runs → writes to `planner-output.md` → registered as {planner_output}
+Agent B runs → receives {planner_output} → builds on Agent A's work
+```
+
+Step 04 collects input/output definitions. Step 05 validates placeholders are properly configured before generating the workflow.
+
 ## Key Rules
 
 1. **Never skip steps** - Each step builds on previous (5 steps total), prompts are injected sequentially
@@ -233,6 +266,7 @@ When user chooses to brainstorm in Step 01, select techniques from this list bas
 7. **Append to plan file immediately** - When user confirms, append data to plan file right away
 8. **Use TodoWrite** - Track progress through steps with the todo list
 9. **Guide user to correct step** - You don't have full context until you reach the right step. If user asks about something that belongs to a later step, guide them to proceed step-by-step. Process selected conditions in order. Example: if user selected `agents` + `prompts` and asks about prompts, say "Before we talk about prompts, let's handle agents first since it's earlier in your journey. Press Enter to proceed to agents."
+10. **Define input/output for agents** - Each agent needs to know where its context comes from (input) and what it produces (output). This enables placeholder chaining between agents.
 
 ## Workflow Plan File