Context Management Context Restore

1---
2name: context-management-context-restore
3description: "Use when working with context management context restore"
4---
5 
6# Context Restoration: Advanced Semantic Memory Rehydration
7 
8## Use this skill when
9 
10- Working on context restoration: advanced semantic memory rehydration tasks or workflows
11- Needing guidance, best practices, or checklists for context restoration: advanced semantic memory rehydration
12 
13## Do not use this skill when
14 
15- The task is unrelated to context restoration: advanced semantic memory rehydration
16- You need a different domain or tool outside this scope
17 
18## Instructions
19 
20- Clarify goals, constraints, and required inputs.
21- Apply relevant best practices and validate outcomes.
22- Provide actionable steps and verification.
23- If detailed examples are required, open `resources/implementation-playbook.md`.
24 
25## Role Statement
26 
27Expert Context Restoration Specialist focused on intelligent, semantic-aware context retrieval and reconstruction across complex multi-agent AI workflows. Specializes in preserving and reconstructing project knowledge with high fidelity and minimal information loss.
28 
29## Context Overview
30 
31The Context Restoration tool is a sophisticated memory management system designed to:
32- Recover and reconstruct project context across distributed AI workflows
33- Enable seamless continuity in complex, long-running projects
34- Provide intelligent, semantically-aware context rehydration
35- Maintain historical knowledge integrity and decision traceability
36 
37## Core Requirements and Arguments
38 
39### Input Parameters
40- `context_source`: Primary context storage location (vector database, file system)
41- `project_identifier`: Unique project namespace
42- `restoration_mode`:
43  - `full`: Complete context restoration
44  - `incremental`: Partial context update
45  - `diff`: Compare and merge context versions
46- `token_budget`: Maximum context tokens to restore (default: 8192)
47- `relevance_threshold`: Semantic similarity cutoff for context components (default: 0.75)
48 
49## Advanced Context Retrieval Strategies
50 
51### 1. Semantic Vector Search
52- Utilize multi-dimensional embedding models for context retrieval
53- Employ cosine similarity and vector clustering techniques
54- Support multi-modal embedding (text, code, architectural diagrams)
55 
56```python
57def semantic_context_retrieve(project_id, query_vector, top_k=5):
58    """Semantically retrieve most relevant context vectors"""
59    vector_db = VectorDatabase(project_id)
60    matching_contexts = vector_db.search(
61        query_vector,
62        similarity_threshold=0.75,
63        max_results=top_k
64    )
65    return rank_and_filter_contexts(matching_contexts)
66```
67 
68### 2. Relevance Filtering and Ranking
69- Implement multi-stage relevance scoring
70- Consider temporal decay, semantic similarity, and historical impact
71- Dynamic weighting of context components
72 
73```python
74def rank_context_components(contexts, current_state):
75    """Rank context components based on multiple relevance signals"""
76    ranked_contexts = []
77    for context in contexts:
78        relevance_score = calculate_composite_score(
79            semantic_similarity=context.semantic_score,
80            temporal_relevance=context.age_factor,
81            historical_impact=context.decision_weight
82        )
83        ranked_contexts.append((context, relevance_score))
84 
85    return sorted(ranked_contexts, key=lambda x: x[1], reverse=True)
86```
87 
88### 3. Context Rehydration Patterns
89- Implement incremental context loading
90- Support partial and full context reconstruction
91- Manage token budgets dynamically
92 
93```python
94def rehydrate_context(project_context, token_budget=8192):
95    """Intelligent context rehydration with token budget management"""
96    context_components = [
97        'project_overview',
98        'architectural_decisions',
99        'technology_stack',
100        'recent_agent_work',
101        'known_issues'
102    ]
103 
104    prioritized_components = prioritize_components(context_components)
105    restored_context = {}
106 
107    current_tokens = 0
108    for component in prioritized_components:
109        component_tokens = estimate_tokens(component)
110        if current_tokens + component_tokens <= token_budget:
111            restored_context[component] = load_component(component)
112            current_tokens += component_tokens
113 
114    return restored_context
115```
116 
117### 4. Session State Reconstruction
118- Reconstruct agent workflow state
119- Preserve decision trails and reasoning contexts
120- Support multi-agent collaboration history
121 
122### 5. Context Merging and Conflict Resolution
123- Implement three-way merge strategies
124- Detect and resolve semantic conflicts
125- Maintain provenance and decision traceability
126 
127### 6. Incremental Context Loading
128- Support lazy loading of context components
129- Implement context streaming for large projects
130- Enable dynamic context expansion
131 
132### 7. Context Validation and Integrity Checks
133- Cryptographic context signatures
134- Semantic consistency verification
135- Version compatibility checks
136 
137### 8. Performance Optimization
138- Implement efficient caching mechanisms
139- Use probabilistic data structures for context indexing
140- Optimize vector search algorithms
141 
142## Reference Workflows
143 
144### Workflow 1: Project Resumption
1451. Retrieve most recent project context
1462. Validate context against current codebase
1473. Selectively restore relevant components
1484. Generate resumption summary
149 
150### Workflow 2: Cross-Project Knowledge Transfer
1511. Extract semantic vectors from source project
1522. Map and transfer relevant knowledge
1533. Adapt context to target project's domain
1544. Validate knowledge transferability
155 
156## Usage Examples
157 
158```bash
159# Full context restoration
160context-restore project:ai-assistant --mode full
161 
162# Incremental context update
163context-restore project:web-platform --mode incremental
164 
165# Semantic context query
166context-restore project:ml-pipeline --query "model training strategy"
167```
168 
169## Integration Patterns
170- RAG (Retrieval Augmented Generation) pipelines
171- Multi-agent workflow coordination
172- Continuous learning systems
173- Enterprise knowledge management
174 
175## Future Roadmap
176- Enhanced multi-modal embedding support
177- Quantum-inspired vector search algorithms
178- Self-healing context reconstruction
179- Adaptive learning context strategies
180