| name | mega-context-manager |
| description | Manages large context windows via chunking, summarization, and retrieval. Us efor long documents or user says manage context, too much context. Use when theuser needs this capability. Triggers: manage context, too much context, c hunkdocument, context budget.
|
| version | 1.1.0 |
| author | Stijnman |
| license | MIT |
| metadata | {"grok":{"tags":["manage context","too much context","chunk document","context budget"],"related_skills":["semantic-memory-manager","predictive-cache-manager","persistent-memory-bridge"]}} |
| compatibility | Grok agent; optional MCP and shell access |
Mega Context Manager
When to Use
- User says manage context or task matches this capability
- User says too much context or task matches this capability
- User says chunk document or task matches this capability
- User says context budget or task matches this capability
Workflow
- Measure token estimate of inputs.
- If over budget: chunk, summarize chunks, index.
- Retrieve relevant chunks per subtask only.
- Drop stale chunks after task completion.
Integrations
semantic-memory-managerpredictive-cache-managerpersistent-memory-bridge
Error Handling
| Failure | Response |
|---|
| Retrieval miss | Expand query; include adjacent chunks. |
Gotchas
- Summaries lose detail; keep raw chunks for citation.
Example
Input: User request matching triggers above.
Output: Structured result per workflow with integrations invoked as needed.
