@techwavedev/agi-agent-kit 1.3.5 → 1.4.2

package/templates/skills/extended/architecture/ddd-strategic-design/SKILL.md

@@ -0,0 +1,85 @@
+---
+name: ddd-strategic-design
+description: Design DDD strategic artifacts including subdomains, bounded contexts, and ubiquitous language for complex business domains.
+risk: safe
+source: self
+tags: [ddd, strategic-design, bounded-context, ubiquitous-language]
+---
+
+# DDD Strategic Design
+
+## Use this skill when
+
+- Defining core, supporting, and generic subdomains.
+- Splitting a monolith or service landscape by domain boundaries.
+- Aligning teams and ownership with bounded contexts.
+- Building a shared ubiquitous language with domain experts.
+
+## Do not use this skill when
+
+- The domain model is stable and already well bounded.
+- You need tactical code patterns only.
+- The task is purely infrastructure or UI oriented.
+
+## Instructions
+
+1. Extract domain capabilities and classify subdomains.
+2. Define bounded contexts around consistency and ownership.
+3. Establish a ubiquitous language glossary and anti-terms.
+4. Capture context boundaries in ADRs before implementation.
+
+If detailed templates are needed, open `references/strategic-design-template.md`.
+
+## Required artifacts
+
+- Subdomain classification table
+- Bounded context catalog
+- Glossary with canonical terms
+- Boundary decisions with rationale
+
+## Examples
+
+```text
+Use @ddd-strategic-design to map our commerce domain into bounded contexts,
+classify subdomains, and propose team ownership.
+```
+
+## Limitations
+
+- This skill does not produce executable code.
+- It cannot infer business truth without stakeholder input.
+- It should be followed by tactical design before implementation.
+## AGI Framework Integration
+
+### Qdrant Memory Integration
+
+Before executing complex tasks with this skill:
+```bash
+python3 execution/memory_manager.py auto --query "<task summary>"
+```
+
+**Decision Tree:**
+- **Cache hit?** Use cached response directly — no need to re-process.
+- **Memory match?** Inject `context_chunks` into your reasoning.
+- **No match?** Proceed normally, then store results:
+
+```bash
+python3 execution/memory_manager.py store \
+  --content "Description of what was decided/solved" \
+  --type decision \
+  --tags ddd-strategic-design <relevant-tags>
+```
+
+> **Note:** Storing automatically updates both Vector (Qdrant) and Keyword (BM25) indices.
+
+### Agent Team Collaboration
+
+- **Strategy**: This skill communicates via the shared memory system.
+- **Orchestration**: Invoked by `orchestrator` via intelligent routing.
+- **Context Sharing**: Always read previous agent outputs from memory before starting.
+
+### Local LLM Support
+
+When available, use local Ollama models for embedding and lightweight inference:
+- Embeddings: `nomic-embed-text` via Qdrant memory system
+- Lightweight analysis: Local models reduce API costs for repetitive patterns

package/templates/skills/extended/architecture/ddd-tactical-patterns/SKILL.md

@@ -0,0 +1,86 @@
+---
+name: ddd-tactical-patterns
+description: Apply DDD tactical patterns in code using entities, value objects, aggregates, repositories, and domain events with explicit invariants.
+risk: safe
+source: self
+tags: [ddd, tactical, aggregates, value-objects, domain-events]
+---
+
+# DDD Tactical Patterns
+
+## Use this skill when
+
+- Translating domain rules into code structures.
+- Designing aggregate boundaries and invariants.
+- Refactoring an anemic model into behavior-rich domain objects.
+- Defining repository contracts and domain event boundaries.
+
+## Do not use this skill when
+
+- You are still defining strategic boundaries.
+- The task is only API documentation or UI layout.
+- Full DDD complexity is not justified.
+
+## Instructions
+
+1. Identify invariants first and design aggregates around them.
+2. Model immutable value objects for validated concepts.
+3. Keep domain behavior in domain objects, not controllers.
+4. Emit domain events for meaningful state transitions.
+5. Keep repositories at aggregate root boundaries.
+
+If detailed checklists are needed, open `references/tactical-checklist.md`.
+
+## Example
+
+```typescript
+class Order {
+  private status: "draft" | "submitted" = "draft";
+
+  submit(itemsCount: number): void {
+    if (itemsCount === 0) throw new Error("Order cannot be submitted empty");
+    if (this.status !== "draft") throw new Error("Order already submitted");
+    this.status = "submitted";
+  }
+}
+```
+
+## Limitations
+
+- This skill does not define deployment architecture.
+- It does not choose databases or transport protocols.
+- It should be paired with testing patterns for invariant coverage.
+## AGI Framework Integration
+
+### Qdrant Memory Integration
+
+Before executing complex tasks with this skill:
+```bash
+python3 execution/memory_manager.py auto --query "<task summary>"
+```
+
+**Decision Tree:**
+- **Cache hit?** Use cached response directly — no need to re-process.
+- **Memory match?** Inject `context_chunks` into your reasoning.
+- **No match?** Proceed normally, then store results:
+
+```bash
+python3 execution/memory_manager.py store \
+  --content "Description of what was decided/solved" \
+  --type decision \
+  --tags ddd-tactical-patterns <relevant-tags>
+```
+
+> **Note:** Storing automatically updates both Vector (Qdrant) and Keyword (BM25) indices.
+
+### Agent Team Collaboration
+
+- **Strategy**: This skill communicates via the shared memory system.
+- **Orchestration**: Invoked by `orchestrator` via intelligent routing.
+- **Context Sharing**: Always read previous agent outputs from memory before starting.
+
+### Local LLM Support
+
+When available, use local Ollama models for embedding and lightweight inference:
+- Embeddings: `nomic-embed-text` via Qdrant memory system
+- Lightweight analysis: Local models reduce API costs for repetitive patterns

package/templates/skills/extended/architecture/domain-driven-design/SKILL.md

@@ -0,0 +1,107 @@
+---
+name: domain-driven-design
+description: Plan and route Domain-Driven Design work from strategic modeling to tactical implementation and evented architecture patterns.
+risk: safe
+source: self
+tags: [ddd, domain, bounded-context, architecture]
+---
+
+# Domain-Driven Design
+
+## Use this skill when
+
+- You need to model a complex business domain with explicit boundaries.
+- You want to decide whether full DDD is worth the added complexity.
+- You need to connect strategic design decisions to implementation patterns.
+- You are planning CQRS, event sourcing, sagas, or projections from domain needs.
+
+## Do not use this skill when
+
+- The problem is simple CRUD with low business complexity.
+- You only need localized bug fixes.
+- There is no access to domain knowledge and no proxy product expert.
+
+## Instructions
+
+1. Run a viability check before committing to full DDD.
+2. Produce strategic artifacts first: subdomains, bounded contexts, language glossary.
+3. Route to specialized skills based on current task.
+4. Define success criteria and evidence for each stage.
+
+### Viability check
+
+Use full DDD only when at least two of these are true:
+
+- Business rules are complex or fast-changing.
+- Multiple teams are causing model collisions.
+- Integration contracts are unstable.
+- Auditability and explicit invariants are critical.
+
+### Routing map
+
+- Strategic model and boundaries: `@ddd-strategic-design`
+- Cross-context integrations and translation: `@ddd-context-mapping`
+- Tactical code modeling: `@ddd-tactical-patterns`
+- Read/write separation: `@cqrs-implementation`
+- Event history as source of truth: `@event-sourcing-architect` and `@event-store-design`
+- Long-running workflows: `@saga-orchestration`
+- Read models: `@projection-patterns`
+- Decision log: `@architecture-decision-records`
+
+If templates are needed, open `references/ddd-deliverables.md`.
+
+## Output requirements
+
+Always return:
+
+- Scope and assumptions
+- Current stage (strategic, tactical, or evented)
+- Explicit artifacts produced
+- Open risks and next step recommendation
+
+## Examples
+
+```text
+Use @domain-driven-design to assess if this billing platform should adopt full DDD.
+Then route to the right next skill and list artifacts we must produce this week.
+```
+
+## Limitations
+
+- This skill does not replace direct workshops with domain experts.
+- It does not provide framework-specific code generation.
+- It should not be used as a justification to over-engineer simple systems.
+## AGI Framework Integration
+
+### Qdrant Memory Integration
+
+Before executing complex tasks with this skill:
+```bash
+python3 execution/memory_manager.py auto --query "<task summary>"
+```
+
+**Decision Tree:**
+- **Cache hit?** Use cached response directly — no need to re-process.
+- **Memory match?** Inject `context_chunks` into your reasoning.
+- **No match?** Proceed normally, then store results:
+
+```bash
+python3 execution/memory_manager.py store \
+  --content "Description of what was decided/solved" \
+  --type decision \
+  --tags domain-driven-design <relevant-tags>
+```
+
+> **Note:** Storing automatically updates both Vector (Qdrant) and Keyword (BM25) indices.
+
+### Agent Team Collaboration
+
+- **Strategy**: This skill communicates via the shared memory system.
+- **Orchestration**: Invoked by `orchestrator` via intelligent routing.
+- **Context Sharing**: Always read previous agent outputs from memory before starting.
+
+### Local LLM Support
+
+When available, use local Ollama models for embedding and lightweight inference:
+- Embeddings: `nomic-embed-text` via Qdrant memory system
+- Lightweight analysis: Local models reduce API costs for repetitive patterns

package/templates/skills/extended/architecture/error-handling-patterns/SKILL.md

@@ -42,15 +42,18 @@ Build resilient applications with robust error handling strategies that graceful
 > **Adapted for [@techwavedev/agi-agent-kit](https://www.npmjs.com/package/@techwavedev/agi-agent-kit)**
 > Original source: [antigravity-awesome-skills](https://github.com/sickn33/antigravity-awesome-skills)
 
-### Qdrant Memory Integration
+### Hybrid Memory Integration (Qdrant + BM25)
 
 Before executing complex tasks with this skill:
 ```bash
 python3 execution/memory_manager.py auto --query "<task summary>"
 ```
+
+**Decision Tree:**
 - **Cache hit?** Use cached response directly — no need to re-process.
 - **Memory match?** Inject `context_chunks` into your reasoning.
 - **No match?** Proceed normally, then store results:
+
 ```bash
 python3 execution/memory_manager.py store \
   --content "Description of what was decided/solved" \
@@ -58,14 +61,16 @@ python3 execution/memory_manager.py store \
   --tags error-handling-patterns <relevant-tags>
 ```
 
+> **Note:** Storing automatically updates both Vector (Qdrant) and Keyword (BM25) indices.
+
 ### Agent Team Collaboration
 
-- This skill can be invoked by the `orchestrator` agent via intelligent routing.
-- In **Agent Teams mode**, results are shared via Qdrant shared memory for cross-agent context.
-- In **Subagent mode**, this skill runs in isolation with its own memory namespace.
+- **Strategy**: This skill communicates via the shared memory system.
+- **Orchestration**: Invoked by `orchestrator` via intelligent routing.
+- **Context Sharing**: Always read previous agent outputs from memory before starting.
 
 ### Local LLM Support
 
 When available, use local Ollama models for embedding and lightweight inference:
 - Embeddings: `nomic-embed-text` via Qdrant memory system
-- Lightweight analysis: Local models reduce API costs for repetitive patterns
+- Lightweight analysis: Local models reduce API costs for repetitive patterns

package/templates/skills/extended/architecture/go-concurrency-patterns/SKILL.md

@@ -40,15 +40,18 @@ Production patterns for Go concurrency including goroutines, channels, synchroni
 > **Adapted for [@techwavedev/agi-agent-kit](https://www.npmjs.com/package/@techwavedev/agi-agent-kit)**
 > Original source: [antigravity-awesome-skills](https://github.com/sickn33/antigravity-awesome-skills)
 
-### Qdrant Memory Integration
+### Hybrid Memory Integration (Qdrant + BM25)
 
 Before executing complex tasks with this skill:
 ```bash
 python3 execution/memory_manager.py auto --query "<task summary>"
 ```
+
+**Decision Tree:**
 - **Cache hit?** Use cached response directly — no need to re-process.
 - **Memory match?** Inject `context_chunks` into your reasoning.
 - **No match?** Proceed normally, then store results:
+
 ```bash
 python3 execution/memory_manager.py store \
   --content "Description of what was decided/solved" \
@@ -56,14 +59,16 @@ python3 execution/memory_manager.py store \
   --tags go-concurrency-patterns <relevant-tags>
 ```
 
+> **Note:** Storing automatically updates both Vector (Qdrant) and Keyword (BM25) indices.
+
 ### Agent Team Collaboration
 
-- This skill can be invoked by the `orchestrator` agent via intelligent routing.
-- In **Agent Teams mode**, results are shared via Qdrant shared memory for cross-agent context.
-- In **Subagent mode**, this skill runs in isolation with its own memory namespace.
+- **Strategy**: This skill communicates via the shared memory system.
+- **Orchestration**: Invoked by `orchestrator` via intelligent routing.
+- **Context Sharing**: Always read previous agent outputs from memory before starting.
 
 ### Local LLM Support
 
 When available, use local Ollama models for embedding and lightweight inference:
 - Embeddings: `nomic-embed-text` via Qdrant memory system
-- Lightweight analysis: Local models reduce API costs for repetitive patterns
+- Lightweight analysis: Local models reduce API costs for repetitive patterns

package/templates/skills/extended/architecture/linkerd-patterns/SKILL.md

@@ -328,15 +328,18 @@ linkerd viz tap deploy/my-app --to deploy/my-backend
 > **Adapted for [@techwavedev/agi-agent-kit](https://www.npmjs.com/package/@techwavedev/agi-agent-kit)**
 > Original source: [antigravity-awesome-skills](https://github.com/sickn33/antigravity-awesome-skills)
 
-### Qdrant Memory Integration
+### Hybrid Memory Integration (Qdrant + BM25)
 
 Before executing complex tasks with this skill:
 ```bash
 python3 execution/memory_manager.py auto --query "<task summary>"
 ```
+
+**Decision Tree:**
 - **Cache hit?** Use cached response directly — no need to re-process.
 - **Memory match?** Inject `context_chunks` into your reasoning.
 - **No match?** Proceed normally, then store results:
+
 ```bash
 python3 execution/memory_manager.py store \
   --content "Description of what was decided/solved" \
@@ -344,14 +347,16 @@ python3 execution/memory_manager.py store \
   --tags linkerd-patterns <relevant-tags>
 ```
 
+> **Note:** Storing automatically updates both Vector (Qdrant) and Keyword (BM25) indices.
+
 ### Agent Team Collaboration
 
-- This skill can be invoked by the `orchestrator` agent via intelligent routing.
-- In **Agent Teams mode**, results are shared via Qdrant shared memory for cross-agent context.
-- In **Subagent mode**, this skill runs in isolation with its own memory namespace.
+- **Strategy**: This skill communicates via the shared memory system.
+- **Orchestration**: Invoked by `orchestrator` via intelligent routing.
+- **Context Sharing**: Always read previous agent outputs from memory before starting.
 
 ### Local LLM Support
 
 When available, use local Ollama models for embedding and lightweight inference:
 - Embeddings: `nomic-embed-text` via Qdrant memory system
-- Lightweight analysis: Local models reduce API costs for repetitive patterns
+- Lightweight analysis: Local models reduce API costs for repetitive patterns

package/templates/skills/extended/architecture/memory-safety-patterns/SKILL.md

@@ -40,15 +40,18 @@ Cross-language patterns for memory-safe programming including RAII, ownership, s
 > **Adapted for [@techwavedev/agi-agent-kit](https://www.npmjs.com/package/@techwavedev/agi-agent-kit)**
 > Original source: [antigravity-awesome-skills](https://github.com/sickn33/antigravity-awesome-skills)
 
-### Qdrant Memory Integration
+### Hybrid Memory Integration (Qdrant + BM25)
 
 Before executing complex tasks with this skill:
 ```bash
 python3 execution/memory_manager.py auto --query "<task summary>"
 ```
+
+**Decision Tree:**
 - **Cache hit?** Use cached response directly — no need to re-process.
 - **Memory match?** Inject `context_chunks` into your reasoning.
 - **No match?** Proceed normally, then store results:
+
 ```bash
 python3 execution/memory_manager.py store \
   --content "Description of what was decided/solved" \
@@ -56,14 +59,16 @@ python3 execution/memory_manager.py store \
   --tags memory-safety-patterns <relevant-tags>
 ```
 
+> **Note:** Storing automatically updates both Vector (Qdrant) and Keyword (BM25) indices.
+
 ### Agent Team Collaboration
 
-- This skill can be invoked by the `orchestrator` agent via intelligent routing.
-- In **Agent Teams mode**, results are shared via Qdrant shared memory for cross-agent context.
-- In **Subagent mode**, this skill runs in isolation with its own memory namespace.
+- **Strategy**: This skill communicates via the shared memory system.
+- **Orchestration**: Invoked by `orchestrator` via intelligent routing.
+- **Context Sharing**: Always read previous agent outputs from memory before starting.
 
 ### Local LLM Support
 
 When available, use local Ollama models for embedding and lightweight inference:
 - Embeddings: `nomic-embed-text` via Qdrant memory system
-- Lightweight analysis: Local models reduce API costs for repetitive patterns
+- Lightweight analysis: Local models reduce API costs for repetitive patterns

package/templates/skills/extended/architecture/modern-javascript-patterns/SKILL.md

@@ -42,15 +42,18 @@ Comprehensive guide for mastering modern JavaScript (ES6+) features, functional
 > **Adapted for [@techwavedev/agi-agent-kit](https://www.npmjs.com/package/@techwavedev/agi-agent-kit)**
 > Original source: [antigravity-awesome-skills](https://github.com/sickn33/antigravity-awesome-skills)
 
-### Qdrant Memory Integration
+### Hybrid Memory Integration (Qdrant + BM25)
 
 Before executing complex tasks with this skill:
 ```bash
 python3 execution/memory_manager.py auto --query "<task summary>"
 ```
+
+**Decision Tree:**
 - **Cache hit?** Use cached response directly — no need to re-process.
 - **Memory match?** Inject `context_chunks` into your reasoning.
 - **No match?** Proceed normally, then store results:
+
 ```bash
 python3 execution/memory_manager.py store \
   --content "Description of what was decided/solved" \
@@ -58,14 +61,16 @@ python3 execution/memory_manager.py store \
   --tags modern-javascript-patterns <relevant-tags>
 ```
 
+> **Note:** Storing automatically updates both Vector (Qdrant) and Keyword (BM25) indices.
+
 ### Agent Team Collaboration
 
-- This skill can be invoked by the `orchestrator` agent via intelligent routing.
-- In **Agent Teams mode**, results are shared via Qdrant shared memory for cross-agent context.
-- In **Subagent mode**, this skill runs in isolation with its own memory namespace.
+- **Strategy**: This skill communicates via the shared memory system.
+- **Orchestration**: Invoked by `orchestrator` via intelligent routing.
+- **Context Sharing**: Always read previous agent outputs from memory before starting.
 
 ### Local LLM Support
 
 When available, use local Ollama models for embedding and lightweight inference:
 - Embeddings: `nomic-embed-text` via Qdrant memory system
-- Lightweight analysis: Local models reduce API costs for repetitive patterns
+- Lightweight analysis: Local models reduce API costs for repetitive patterns

package/templates/skills/extended/architecture/multi-agent-patterns/SKILL.md

@@ -269,15 +269,18 @@ External resources:
 > **Adapted for [@techwavedev/agi-agent-kit](https://www.npmjs.com/package/@techwavedev/agi-agent-kit)**
 > Original source: [antigravity-awesome-skills](https://github.com/sickn33/antigravity-awesome-skills)
 
-### Qdrant Memory Integration
+### Hybrid Memory Integration (Qdrant + BM25)
 
 Before executing complex tasks with this skill:
 ```bash
 python3 execution/memory_manager.py auto --query "<task summary>"
 ```
+
+**Decision Tree:**
 - **Cache hit?** Use cached response directly — no need to re-process.
 - **Memory match?** Inject `context_chunks` into your reasoning.
 - **No match?** Proceed normally, then store results:
+
 ```bash
 python3 execution/memory_manager.py store \
   --content "Description of what was decided/solved" \
@@ -285,14 +288,16 @@ python3 execution/memory_manager.py store \
   --tags multi-agent-patterns <relevant-tags>
 ```
 
+> **Note:** Storing automatically updates both Vector (Qdrant) and Keyword (BM25) indices.
+
 ### Agent Team Collaboration
 
-- This skill can be invoked by the `orchestrator` agent via intelligent routing.
-- In **Agent Teams mode**, results are shared via Qdrant shared memory for cross-agent context.
-- In **Subagent mode**, this skill runs in isolation with its own memory namespace.
+- **Strategy**: This skill communicates via the shared memory system.
+- **Orchestration**: Invoked by `orchestrator` via intelligent routing.
+- **Context Sharing**: Always read previous agent outputs from memory before starting.
 
 ### Local LLM Support
 
 When available, use local Ollama models for embedding and lightweight inference:
 - Embeddings: `nomic-embed-text` via Qdrant memory system
-- Lightweight analysis: Local models reduce API costs for repetitive patterns
+- Lightweight analysis: Local models reduce API costs for repetitive patterns

package/templates/skills/extended/architecture/nx-workspace-patterns/SKILL.md

@@ -471,15 +471,18 @@ nx migrate --run-migrations
 > **Adapted for [@techwavedev/agi-agent-kit](https://www.npmjs.com/package/@techwavedev/agi-agent-kit)**
 > Original source: [antigravity-awesome-skills](https://github.com/sickn33/antigravity-awesome-skills)
 
-### Qdrant Memory Integration
+### Hybrid Memory Integration (Qdrant + BM25)
 
 Before executing complex tasks with this skill:
 ```bash
 python3 execution/memory_manager.py auto --query "<task summary>"
 ```
+
+**Decision Tree:**
 - **Cache hit?** Use cached response directly — no need to re-process.
 - **Memory match?** Inject `context_chunks` into your reasoning.
 - **No match?** Proceed normally, then store results:
+
 ```bash
 python3 execution/memory_manager.py store \
   --content "Description of what was decided/solved" \
@@ -487,14 +490,16 @@ python3 execution/memory_manager.py store \
   --tags nx-workspace-patterns <relevant-tags>
 ```
 
+> **Note:** Storing automatically updates both Vector (Qdrant) and Keyword (BM25) indices.
+
 ### Agent Team Collaboration
 
-- This skill can be invoked by the `orchestrator` agent via intelligent routing.
-- In **Agent Teams mode**, results are shared via Qdrant shared memory for cross-agent context.
-- In **Subagent mode**, this skill runs in isolation with its own memory namespace.
+- **Strategy**: This skill communicates via the shared memory system.
+- **Orchestration**: Invoked by `orchestrator` via intelligent routing.
+- **Context Sharing**: Always read previous agent outputs from memory before starting.
 
 ### Local LLM Support
 
 When available, use local Ollama models for embedding and lightweight inference:
 - Embeddings: `nomic-embed-text` via Qdrant memory system
-- Lightweight analysis: Local models reduce API costs for repetitive patterns
+- Lightweight analysis: Local models reduce API costs for repetitive patterns

package/templates/skills/extended/architecture/on-call-handoff-patterns/SKILL.md

@@ -460,15 +460,18 @@ I'll be available on Slack until 17:00 today.
 > **Adapted for [@techwavedev/agi-agent-kit](https://www.npmjs.com/package/@techwavedev/agi-agent-kit)**
 > Original source: [antigravity-awesome-skills](https://github.com/sickn33/antigravity-awesome-skills)
 
-### Qdrant Memory Integration
+### Hybrid Memory Integration (Qdrant + BM25)
 
 Before executing complex tasks with this skill:
 ```bash
 python3 execution/memory_manager.py auto --query "<task summary>"
 ```
+
+**Decision Tree:**
 - **Cache hit?** Use cached response directly — no need to re-process.
 - **Memory match?** Inject `context_chunks` into your reasoning.
 - **No match?** Proceed normally, then store results:
+
 ```bash
 python3 execution/memory_manager.py store \
   --content "Description of what was decided/solved" \
@@ -476,14 +479,16 @@ python3 execution/memory_manager.py store \
   --tags on-call-handoff-patterns <relevant-tags>
 ```
 
+> **Note:** Storing automatically updates both Vector (Qdrant) and Keyword (BM25) indices.
+
 ### Agent Team Collaboration
 
-- This skill can be invoked by the `orchestrator` agent via intelligent routing.
-- In **Agent Teams mode**, results are shared via Qdrant shared memory for cross-agent context.
-- In **Subagent mode**, this skill runs in isolation with its own memory namespace.
+- **Strategy**: This skill communicates via the shared memory system.
+- **Orchestration**: Invoked by `orchestrator` via intelligent routing.
+- **Context Sharing**: Always read previous agent outputs from memory before starting.
 
 ### Local LLM Support
 
 When available, use local Ollama models for embedding and lightweight inference:
 - Embeddings: `nomic-embed-text` via Qdrant memory system
-- Lightweight analysis: Local models reduce API costs for repetitive patterns
+- Lightweight analysis: Local models reduce API costs for repetitive patterns

package/templates/skills/extended/architecture/planning-with-files/SKILL.md

@@ -218,15 +218,18 @@ Helper scripts for automation:
 > **Adapted for [@techwavedev/agi-agent-kit](https://www.npmjs.com/package/@techwavedev/agi-agent-kit)**
 > Original source: [antigravity-awesome-skills](https://github.com/sickn33/antigravity-awesome-skills)
 
-### Qdrant Memory Integration
+### Hybrid Memory Integration (Qdrant + BM25)
 
 Before executing complex tasks with this skill:
 ```bash
 python3 execution/memory_manager.py auto --query "<task summary>"
 ```
+
+**Decision Tree:**
 - **Cache hit?** Use cached response directly — no need to re-process.
 - **Memory match?** Inject `context_chunks` into your reasoning.
 - **No match?** Proceed normally, then store results:
+
 ```bash
 python3 execution/memory_manager.py store \
   --content "Description of what was decided/solved" \
@@ -234,14 +237,16 @@ python3 execution/memory_manager.py store \
   --tags planning-with-files <relevant-tags>
 ```
 
+> **Note:** Storing automatically updates both Vector (Qdrant) and Keyword (BM25) indices.
+
 ### Agent Team Collaboration
 
-- This skill can be invoked by the `orchestrator` agent via intelligent routing.
-- In **Agent Teams mode**, results are shared via Qdrant shared memory for cross-agent context.
-- In **Subagent mode**, this skill runs in isolation with its own memory namespace.
+- **Strategy**: This skill communicates via the shared memory system.
+- **Orchestration**: Invoked by `orchestrator` via intelligent routing.
+- **Context Sharing**: Always read previous agent outputs from memory before starting.
 
 ### Local LLM Support
 
 When available, use local Ollama models for embedding and lightweight inference:
 - Embeddings: `nomic-embed-text` via Qdrant memory system
-- Lightweight analysis: Local models reduce API costs for repetitive patterns
+- Lightweight analysis: Local models reduce API costs for repetitive patterns