OPEN SPEED OPTIMISATION LOOP

A reusable speed optimization framework for AI agents. Proven patterns that turn any codebase faster, one loop at a time.

npx skills add https://github.com/toomingos/open_speed_optimization_loop --skill speed-loop

Works with

& more

Setup

Speed Loop

CONTINUOUS

Sacred Rules

Output Integrity

Same inputs, same outputs.

Verification Checkpoints

No merge without evidence.

Correctness Over Speed

Wrong results = failed system.

Test Isolation

Verify before integration.

One Per Iteration

Isolate causation.

Document Everything

Hypothesis, measurements, decision.

Rollback Readiness

Every change is reversible.

Case Study

81%faster

1,829ms→352ms16 loops →

01UnderstandMap & measure

02DecomposePrioritize candidates

03AnalyzeDeep-dive

04ImplementBuild isolated

05VerifyCompare & prove

06IntegrateMerge & archive

CASE STUDY

16 optimization loops on a user journey mapping algorithm. Baseline: 1,829ms.

Approved

Rejected

Skipped

Read the docs, explore the methodology, and start optimizing.