README.md
Core Principle
Parallelise execution, never parallelise truth.
System of Record for Reasoning
SoRR
Context. Constraints. Decisions. Traceability.
A structured operating layer for human and AI collaboration. SoRR keeps projects grounded in explicit truth, reduces execution drift, and makes decisions, evidence, and constraints durable across every session.
README.mdlive
CURRENT_STATE.mdlive
DECISIONS.mdlive
FINDINGS.mdlive
AI_CONSTRAINTS.mdlive
TESTING.mdlive
Why SoRR
Most delivery problems are not caused by a lack of tools. They come from missing context, hidden assumptions, undocumented decisions, and agents acting without clear boundaries.
Structured Context
Every project starts with a clear operating context: scope, constraints, goals, architecture, and the current truth state.
Reasoning Traceability
Important decisions, trade-offs, assumptions, and findings are recorded so work can be understood, reviewed, and continued safely.
Execution Guardrails
Agents and humans work within explicit constraints, reducing drift, rework, hallucination, and undocumented changes.
Continuous Learning
New findings flow back into the record so future work starts smarter, faster, and with less repeated context loading.
Meet the Record
A practical SoRR implementation is made of small, durable documents that define how work should happen and what is true right now.
Core Document
README
Project purpose, current state, operating model, and how to enter the workspace safely.
Core Document
DECISIONS.md
Key trade-offs, architectural choices, and why specific paths were taken.
Core Document
FINDINGS.md
Observed behaviours, experiments, insights, and evidence gathered during delivery.
Core Document
AI_CONSTRAINTS.md
Rules for agent behaviour, allowed actions, hard stops, and truth-handling requirements.
Core Document
CURRENT_STATE.md
What is actually built, what is missing, what is blocked, and what should happen next.
Core Document
TESTING.md
How validation works, what has been tested, what failed, and what remains unverified.
Example Structure
Excerpts from a sample SoRR docs set. This is intentionally selective rather than a full dump.
AI_CONSTRAINTS.md
Rules
Do not act without reading CURRENT_STATE.md. Do not invent missing context. Stop if uncertainty is high.
CURRENT_STATE.md
Required Sections
What Exists, In Progress, Blockers, and Next Actions are tracked so work starts from current reality.
DECISIONS.md
Log Format
Date | Decision | Rationale | Impact
How It Works
1
Define the operating context
Capture project scope, objectives, constraints, current status, and known truths before work begins.
2
Document decisions and findings
Record meaningful choices, learnings, dead ends, and evidence as the project evolves.
3
Constrain execution
Give agents and collaborators clear rules for what they can do, what they must read, and when they should stop.
4
Update the record continuously
Feed new truth back into the system so future work starts from reality, not memory or guesswork.
5
Scale with confidence
Reuse the pattern across products, experiments, and teams without losing consistency or control.
Ready to make reasoning durable?
Use SoRR to ground product work, documentation, AI-assisted delivery, and multi-agent execution in a shared source of truth.
Outcomes
Less drift
Teams and agents stop reinventing the same context on every session.
Cleaner handovers
Work can move between people, tools, and models without losing the thread.
Safer AI execution
Models operate against explicit constraints instead of inferred assumptions.
Audit-ready history
Decisions, evidence, and changes are easier to inspect after the fact.