zlh-grind/README.md

93 lines
2.2 KiB
Markdown

# zlh-grind
GPT Execution Workspace for ZeroLagHub.
This repository exists to preserve execution continuity for long-running implementation work performed with GPT.
It is the **execution workspace** and **working memory layer** for active implementation.
It is **not** the final architecture source of truth.
---
## What This Repo Is For
Use this repository to preserve:
- active execution context
- unfinished work
- implementation findings
- session handovers
- operational snapshots
- chronological session history
The goal is simple: a new session should be able to enter this repository, read a small set of files, and continue work safely.
---
## Front Door
Start here, then continue in this order:
1. `SESSION_START.md`
2. `OPEN_THREADS.md`
3. `CONSTRAINTS.md`
4. `PROJECT_CONTEXT.md` *(if current platform state matters)*
5. `INFRASTRUCTURE.md` *(if VM/IP context matters)*
6. `UPSTREAMS.md` *(if cross-repo context matters)*
Do **not** start by digging through history unless the live docs are insufficient.
---
## Live Working Set
These root files are the live control panel for the repo:
- `SESSION_START.md`
- `OPEN_THREADS.md`
- `CONSTRAINTS.md`
- `PROJECT_CONTEXT.md`
- `INFRASTRUCTURE.md`
- `UPSTREAMS.md`
Keep these current and concise.
---
## History And Deep Context
Use these only when deeper recovery is needed:
- `SESSION_LOG/` — chronological implementation history
- `Session_Summaries/` — session handovers and milestone summaries
- `SCRATCH/` — temporary or investigative notes
- `docs/` — reference and archive material
- `MANAGEMENT/` — non-core planning material
These support the live docs, but they are not the normal starting point for a fresh session.
---
## Repo Boundary
This repository should not become a dumping ground for every possible context artifact.
Do not let root-level docs multiply without a clear reason.
Root docs should answer only one question:
**What must a new session know right now to continue work safely?**
---
## Session Rule
Never assume speculative features are implemented.
When in doubt:
- trust the live working set first
- use history folders for conflict resolution
- update the live docs after meaningful work is completed