This system is currently published for visibility, authorship, and reflective exploration.
It is not open source or commercially licensable at this time.
See LICENSE.md for usage terms.
Contact Peter Salvato directly for consultation, collaboration, or licensing inquiries.
Canonical Public Release: v1.0 (2025-04-08)
Protocol Syntax Version: 3.0
Author: Peter Salvato
© 2025 Peter Salvato. All rights reserved.
All terminology, structure, and logic protected under custom license.
Savepoint is the machete, not the map.
Savepoint.Protocol™ is a symbolic markup system for capturing cognitive turning points—moments when realization clicks, phrasing locks in, or meaning shifts direction.
It’s not memory. It’s authorship infrastructure.
Built for thinkers in wild terrain.
Savepoints are not summaries. Not notes. Not metadata. They are semantic trail markers—a protocol for claiming where your thinking changed.
Modern systems—especially AI—are building passive memory: endless recall, zero perspective.
They remember everything, but not what mattered. They store what you said, but not what changed you.
This is the problem: black box memory. You can search your history but not trace your authorship.
Savepoint.Protocol exists to solve that.
- It captures semantic inflection—when something clicked
- It preserves your phrasing—not just data
- It restores authorship in a sea of machine retention
Whether you’re writing, designing, debugging, or philosophizing: Savepoints let you track when meaning moved.
A Savepoint is:
- A self-authored realization
- Timestamped and structured
- Stored in plaintext, written by hand, or parsed from AI logs
- A system-agnostic semantic artifact
You don’t need a platform. You need a symbol.
Savepoints are declared using a self-closing tag:
<Savepoint
protocol_version:3.0
category:system_logic
function:declaration
timestamp:2025-04-08T15:43:00Z
importance:high
influence:Order of the Ætherwright
# Recursive structures should replace version snapshots wherever drift is likely.
/>
protocol_version: always declared (current = 3.0)category: domain of reflection (e.g.design_note,model_shift)function: purpose (declaration,revision,drift_detected)timestamp: ISO 8601 UTC format#: the actual semantic content
importance: high, medium, lowconfidence: self-assessed certaintyinfluence: source, reference, or attribution
- Writers marking breakthroughs in phrasing or structure
- Designers noting system shifts or metaphor realignment
- Researchers capturing conceptual pivots
- Developers logging architecture changes in thinking
- Philosophers flagging turns in internal logic
A Python CLI tool is included to extract Savepoints from exported ChatGPT logs:
git clone https://github.com/peterSalvato/savepoint-protocol.git
cd savepoint-protocol
python cli/savepoint_split.py conversations.json- Parses logs into Savepoint
.mdentries - Timestamps and Git-prepares each session
- Requires Python 3.7+
Tool: cli/savepoint_split.py
/spec/ – Canonical protocol format and rules
/cli/ – CLI tool (ChatGPT log parser)
/examples/ – Sample Savepoint sessions
/.savepoints/ – Real Savepoint logs
/docs/ – Extended philosophy and integration
Savepoint is not a tool. It’s a symbolic language for tracing what matters.
It is:
- Low-tech
- High-trust
- Stack-agnostic
- Format-degradable
- Legible to LLMs and legible to you
It resists feature creep. It honors inflection over information. It builds cognitive version control—not just note capture.
When the forest regrows, the Savepoints remain.
Savepoint.Protocol is licensed under a custom humanist license.
Use freely. Fork deeply. But preserve authorship and intent.
License: license.md
Peter Salvato
Author of Savepoint.Protocol
System Architect: Order of the Ætherwright