proagents 1.6.17 → 1.6.19

package/.proagents/workflow-modes/entry-modes.md

@@ -55,6 +55,30 @@ The complete 10-phase development workflow for new features.
 └─────────────────────────────────────────────────────────────┘
 ```
 
+### Flow Context Check
+
+Before starting, check `.proagents/worklog/_context.md` for Flow Context:
+
+```
+📋 Checking for prior context...
+
+If Flow Context exists:
+┌─────────────────────────────────────────────────────────┐
+│ 📋 Continuing from previous workflow                    │
+│                                                         │
+│ Last Command: pa:rnd-compare                            │
+│ Decision: Zustand for state management                  │
+│ Context: E-commerce cart state                          │
+│ Research: docs/research/comparisons/state-mgmt.md       │
+│                                                         │
+│ Use this context? (yes/no/modify)                       │
+└─────────────────────────────────────────────────────────┘
+
+If "yes": Pre-populate feature with context data
+If "no": Start fresh, clear Flow Context
+If "modify": Allow user to adjust context
+```
+
 ### Starting Full Workflow
 
 ```bash
@@ -63,6 +87,9 @@ pa:feature start "User Profile Management"
 
 # With options
 pa:feature start "User Profile" --skip-ui --checkpoint=critical
+
+# Continue from RND (auto-reads Flow Context)
+pa:feature start "Cart State" --from-rnd
 ```
 
 ### Checkpoints

package/.proagents/worklog/_context.template.md

@@ -72,6 +72,53 @@ Failing: [Test names or "None"]
 | Open Issues | [N] |
 ```
 
+---
+
+## Flow Context
+<!-- Tracks command chain for seamless transitions between pa: commands -->
+```
+Current Flow: [command chain, e.g., "rnd → feature → plan"]
+Last Command: [last pa: command run]
+Decision: [key decision made, if any]
+Context: [relevant context to carry forward]
+Next Suggested: [suggested next command]
+Data: [JSON data to pass to next command]
+```
+
+### Flow Context Rules
+
+1. **Write after completion** - Each command writes its output to Flow Context
+2. **Read on start** - Commands check Flow Context for relevant prior decisions
+3. **Offer continuation** - If Flow Context has Next Suggested, offer to run it
+4. **Clear on explicit start** - When user starts fresh workflow, clear Flow Context
+5. **Preserve data** - JSON data carries structured info (tech stack, decisions, etc.)
+
+### Example Flow
+
+```markdown
+## Flow Context
+```
+Current Flow: rnd → feature
+Last Command: pa:rnd-compare
+Decision: Zustand for state management
+Context: E-commerce cart with complex state
+Next Suggested: pa:feature "cart state management"
+Data: {"tech": "zustand", "use_case": "cart", "research_doc": "docs/research/comparisons/state-management-2024-03-17.md"}
+```
+```
+
+When pa:feature starts, it reads this and says:
+```
+📋 Continuing from R&D...
+
+You recently decided on Zustand for cart state management.
+Research: docs/research/comparisons/state-management-2024-03-17.md
+
+Start feature with this context? (yes/no/different)
+```
+
+---
+
 ## Auto-Update Rules
 
 1. **Update `Last Updated`** with current timestamp