| name | reverse-engineering-improvement-loop |
| description | Convert completed reverse-engineering work into better future workflows, skills, and panel narratives. Use after meaningful milestones, repeated friction, or major analysis wins. |
| aliases | ["re-learning-loop","binary-analysis-improvement","re-retrospective"] |
| category | Developer Tools |
| tags | ["reverse-engineering","memory","evidence","improvement","orchestration"] |
| version | 1.0.0 |
| resolutionHints | ["improve reverse-engineering workflow over time","capture lessons from a reversing session","update re skills and panel memory"] |
| metadata | {"source":"evokore","derived_from":["panel-of-experts","persistent-narratives","improvement-cycles","session-wrap"]} |
Reverse Engineering Improvement Loop
This skill keeps reverse engineering from resetting to zero every session. It is the slim memory layer: keep only what future runs can reuse.
Run This After
- a major unknown binary onboarding session
- a subsystem naming campaign
- a debugger session that resolved a recurring ambiguity
- a malware triage pass that changed the investigation direction
- any milestone where the same confusion or workaround showed up more than once
Workflow
- Review the evidence bundle, not just the final narrative.
- Extract recurring friction:
- wrong first tool choice
- missing checklist
- repeated breakpoint pattern
- weak handoff artifact
- panel coverage gap
- Convert the friction into a concrete improvement target:
- new skill
- updated workflow
- persona refinement
- new evidence template
- rule or automation candidate
- Keep only reusable conclusions in persistent memory.
- Feed the rest into normal session output and move on.
EVOKORE Hooks to Use
session-replay
evidence-capture
persistent-narratives
improvement-cycles
session-wrap
Good Output
- one durable lesson with proof
- one workflow change or skill candidate
- one memory update that future sessions should inherit
Anti-Pattern
Do not dump full transcripts into memory. If a future analyst cannot act on it quickly, it does not belong in the persistent layer.