The /open workflow starts a new working session with full context from previous work. It loads the project contract, latest session log, backlog, and sync queue — then presents a concise report so the user knows exactly where they left off.
Commands Overview
/open [project-name]Session Open Flow
Project paths → Contract → Ecosystem detection → Rules (Two-Tier) → Session log
→ Brainstorms → Backlog → Plan → Strategy → Roadmap → Sync queue → Git → Report1. Identify Project Paths
Resolves the standard PARA project structure:
Projects/[project-name]/
├── repo/ # Source code (git root)
├── sessions/ # Session logs
├── artifacts/ # Tasks, backlog, plans
├── docs/ # Project documentation
└── project.md # Project contract (YAML)2. Read Project Contract
Reads project.md to extract goal, deadline, status, DoD, and downstream projects.
Ecosystem detection (v1.6.0+): Checks type field — if type: ecosystem, notes satellite list and skips git steps later. If ecosystem field exists on a satellite, notes parent ecosystem for @ prefix resolution.
Strategy context (v1.6.3+): Checks strategy field in project.md. Resolves the path using the Path Resolution Convention (supports @ prefix), and extracts the title and first blockquote (~30 tokens) for the report. Avoids filesystem probing.
2.5. Load Workspace Agent Indices — ALWAYS
Reads .agent/rules.md and .agent/skills.md from the workspace root. The agent memorizes the trigger conditions but does NOT read the actual files yet.
2.6. Load Project Agent Indices — CONDITIONAL
Checks project.md for agent.rules: true and agent.skills: true. If enabled, reads the project-specific agent indices.
✅ Completion Gate (v1.6.3): The agent MUST verify that both workspace indices and enabled project indices have been successfully loaded BEFORE continuing to Step 3. This strict gating prevents missing trigger definitions and ensures proactive bug prevention.
During the session, when an action matches a trigger from either index, the agent loads the corresponding rule file before acting.
3. Find Latest Session
Reads the most recent session log to understand what was done and what’s pending.
3.5. Check Pending Brainstorms (v1.6.1)
Checks for brainstorms with pending decisions. If found, displays topic and date in the report for follow-up.
4. Read Task Context (Token Optimized)
Token optimization: Reads backlog summary (~10 lines) + hot lane. Never reads full backlog.
4a. Backlog Summary — Always reads the 📊 Summary table and top active items (grep).
4b. Hot Lane — Reads sprint-current.md if it exists (small file, ~50-100 tokens). If not exists, reports 🔥 Hot Lane: empty. This file holds ad-hoc quick tasks from previous sessions.
5. Read Plan (if active)
If project.md has an active_plan field, reads phase headers and the backlog-to-phase mapping to identify the current phase. Skipped if no plan exists.
Cross-project plans (v1.6.0+): If active_plan starts with @, resolves the path across project boundaries (e.g., @my-ecosystem/plans/xxx.md → Projects/my-ecosystem/artifacts/plans/xxx.md).
5.5. Roadmap Context Loading (v1.6.3)
Checks the roadmap field in project.md (field-gated). If it exists, resolves the path using the Path Resolution Convention and extracts the Phases Overview table. Globs are no longer used for discovery, ensuring targeted I/O operations.
Strategy cascade detection (D10): If BOTH strategy and roadmap exist and strategy is newer than roadmap, warns: ⚠️ Strategy updated after roadmap. Roadmap may need review.
6. Check Sync Queue (Conditional)
Token optimization: Only reads
SYNC.mdifproject.md(loaded in Step 2) hasdownstreamorupstreamfields. Otherwise skips entirely.
When applicable, reads Areas/Workspace/SYNC.md for pending cross-project notifications targeting this project.
7. Check Git Status
Quick git status --short and latest commit to show working tree state.
Ecosystem projects (v1.6.0+): Skipped entirely for type: ecosystem projects (no repo/ directory).
8. Display Report
🚀 Starting: [Project Name] | 📅 [Date]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
📋 LAST SESSION: [Date] - [Focus]
✅ Completed: [items]
⏳ Pending TODO: [items]
📄 STRATEGY: [one-line summary]
🗺️ ROADMAP: [name] ([N] phases: [done]✅ [active]🔨 [planned]📋)
📐 ACTIVE PLAN: [plan-name] Phase [X/Y]
💭 BRAINSTORM PENDING: [topic] (date)
🔔 SYNC QUEUE: [N pending]
📝 BACKLOG SUMMARY: High: N | Medium: N | Low: N
🔥 HOT LANE: [pending quick tasks or "empty"]
🛡️ SAFETY (persist across truncation):
- Git: Do NOT merge/branch/tag without user approval.
- Governance: Do NOT modify kernel/ or .para/.
- Recovery: If rules feel incomplete → re-read .agent/rules.md
💡 SUGGESTED ACTIONS: [priority-based]
❓ What would you like to work on?Priority logic (v1.6.1): Suggested Actions are ordered by: active plan tasks → strategy cascade / sync pending → pending brainstorms → roadmap next phase → hot lane items.
Updated in v1.6.3. Added Completion Gate and field-gated Strategy/Roadmap loading (v1.6.3).
→ Workflow /end — Close session with Hot Lane sync → Learn: Rule Layers & Trigger Index → Learn: Ecosystem Architecture → Learn: Strategy → Plan Flow