SmarterContext
Issue #002 · Sat, Jun 27, 2026
Your AI keeps forgetting what you told it. Learn to fix that. |
Project context vs personal context — keep them separate
Your coding style and your company's rules are two different files. Mixing them is why the AI contradicts itself. |
There are two layers of context and people jam them into one. Personal context is you — your voice, how terse you like answers, your preferred languages — and it follows you everywhere. Project context is this codebase / this client / this campaign and it changes per project. Keep them in separate files and compose them at session start. When something feels universal to you, it goes in personal; when it's true only here, it goes in project. This stops the AI from applying a client's banned-words list to your personal blog. |
▸ The template
# ~/personal.md (you, everywhere)
Voice: direct, dry, no hype.
Defaults: Python > everything, tabs over spaces.
Always: explain tradeoffs before recommending.
# ./project.md (THIS repo only)
Client: Acme Health — HIPAA, no PII in logs ever.
Voice override: formal, third-person, cite sources.
Banned words: "revolutionary", "seamless".
# Session start, paste in order:
# personal.md THEN project.md
# (project overrides personal on conflict) |
▸ Level it up
New to this: Split your one big context into two: a 'me' file and a 'this project' file. Paste both, me first.
Vibe coder: Establish a rule: project context overrides personal context on any conflict, and write that rule at the top of the project file.
Advanced: Store personal context in a synced dotfile and pull project context from the repo itself; a wrapper script concatenates them in precedence order before every agent run. |
Takeaway: Two layers, two files: personal follows you, project overrides it locally. |
|
|
|
Know someone fighting the same context problem?
Forward it. Refer 3 friends and we'll send you our private context-template pack.
Refer & get the template pack → |
|
|
|
|