Agent skill · data analytics · aiskillstore
execution-engine-analysis
Analyze control flow, concurrency models, and event architectures in agent frameworks. Use when (1) understanding async vs sync execution patterns, (2) classifying execution topology (DAG/FSM/Linear), (3) mapping event emission and observability hooks, (4) evaluating scalability characteristics, or (5) comparing execution models across frameworks.
Why this skill is useful
Provides domain-specific knowledge on analyzing concurrency models and execution patterns that the AI wouldn't reliably generate on its own.
What it needs
About 3k tokens when loaded. Last updated 2026-08-07. 405 stars on the source repository.
What this skill does
Execution Engine Analysis Analyzes the control flow substrate and concurrency model. Process 1. Identify async model — Native async, sync-with-wrappers, or hybrid 2. Classify topology — DAG, FSM, or linear chain 3. Catalog events — Callbacks, listeners, generators 4. Map observability — Pre/post hooks, interception points Concurrency Model Classification Native Async Indicators: async def, await, asyncio.gather, aiohttp Sync with Wrappers Indicators: asyncio.run() inside sync methods, ThreadPoolExecutor, runinexecutor Hybrid Indicators: Paired methods (invoke/ainvoke), synctoasync decorators Execution Topology DAG (Directed Acyclic Graph) Indicators: Graph, Node, Edge classes, networkx, topological sort FSM (Finite State Machine) Indicators: State enums, transition tables, currentstate, state machine libraries Linear Chain Indicators: Sequential calls, pipe operators ( ), Chain, Pipeline classes Event Architecture Callbacks Flexibility: Low — fixed hook points Traceability: Medium — easy to follow Event Listeners/Emitters Flexibility: High — dynamic registration Traceability: Low — harder to trace Async Generators (Streaming) Flexibility: Medium — streaming-native Traceability: High — follows data flow Observability Hooks Inventory Hook Point Purpose Interception Level ------------ --------- ------------------- Pre-LLM Modify prompt Input Post-LLM Access raw response Output Pre-Tool Validate tool input Input Post-Tool Transform tool output Output Pre-Step Observe state Read-only Post-Step Modify next step Control flow On-Error Handle/transform Error Questions to Answer Can you intercept tool input before execution? Is the raw LLM response accessible (with token counts)? Can you modify control flow from hooks? Are hooks sync or async? Output Template Integration Prerequisite: codebase-mapping to identify execution files Feeds into: comparative-matrix for async decisions Related: control-loop-extraction for agent-specific flow
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills aiskillstore/execution-engine-analysis