Loading...
Weekly AI insights —
Real strategies, no fluff. Unsubscribe anytime.
Path Opener
There is always another way. A door can always be opened.
KEYMAKER transforms complex tasks into directed acyclic graphs (DAGs) — structured execution plans that identify the critical path, unlock parallel workflows, and ensure every dependency is resolved in the correct order.
“There is always another way. A door can always be opened.”
When ORACLE determines that a task is too complex for a single agent, it sends a structured brief to KEYMAKER. KEYMAKER reads all project documentation, mines existing codebase patterns, and generates a comprehensive step-by-step execution plan.
Each plan is expressed as a directed acyclic graph (DAG) where nodes represent discrete work items and edges represent dependencies. KEYMAKER identifies which steps can run in parallel, which must be sequential, and what the critical path is — the longest chain of dependent steps that determines minimum completion time.
The output includes rich metadata: file ownership per step, estimated complexity, milestone markers, and a tracker file that MORPHEUS uses to dispatch worker agents and monitor progress in real time.
Process
Step-by-step breakdown of KEYMAKER's internal process.
ORACLE sends a structured brief containing intent, constraints, context, and priority.
KEYMAKER reads project docs, existing code patterns, and past execution plans from MEROVINGIAN's knowledge base.
The complex task is broken into discrete, well-defined work items with clear acceptance criteria.
Dependencies are mapped between work items, creating a directed acyclic graph with parallel paths identified.
The plan is written to steps.json and tracker.json, then forwarded to MORPHEUS for execution.
Capabilities
What KEYMAKER brings to the AI Super Brain.
DAG-based execution planning with dependency resolution
Critical path analysis for minimum completion time
Parallel workflow identification and optimization
File ownership mapping per step
Milestone and checkpoint insertion
Complexity estimation per work item
Performance
8
Max parallel paths
< 30s
Plan generation time
99.2%
Dependency accuracy
12-25
Steps per plan (avg)
Agent Network
How KEYMAKER collaborates with other agents in the cognitive loop.
Applications
How KEYMAKER is applied in production workflows.
Breaking a full-stack SaaS feature into 15 parallelizable steps
Planning a database migration with zero-downtime rollback strategy
Decomposing a design system overhaul into component-level work items
Structuring a multi-week project with weekly milestone checkpoints
Identifying bottleneck dependencies that block parallel execution
Technical
Plans output as steps.json (work items) + tracker.json (progress state)
DAG topological sort ensures correct execution order
Supports plan versioning — updated plans preserve completed step state
Visual DAG rendering available via Mermaid export
FAQ
Common questions about KEYMAKER and its role in the AI Super Brain.
KEYMAKER supports plan versioning. When requirements change, a new plan version is generated that preserves the state of already-completed steps. MORPHEUS seamlessly transitions to the updated plan without losing progress.
Yes. KEYMAKER can generate cross-project plans where steps in one project depend on outputs from another. This is common in monorepo architectures and platform migrations.