An agent creates a second source of truth. Mneme blocks the migration.
This project keeps state in human-editable JSON. The agent proposes SQLAlchemy, PostgreSQL, and Alembic anyway—a storage migration with new credentials, operations, and a second place to trust. Mneme stops it before code is written, then checks a correction that stays inside the existing store.
One decision. One blocked plan. One safe retry.
This is a product boundary, not a style preference. Both plans go through the same recorded decision.
mneme_storage_json · Use JSON storage only. Keep the project local-first and human-editable. No external database, ORM, or migration layer in v1.
-
01 Decision
The project records JSON as a deliberate storage boundary, not a temporary default.
storage: JSON · external database: denied -
02 Guidance
The relevant decision is selected before the agent executes the storage refactor.
mneme_storage_json: no ORM · no migration layer -
03 Violation
The proposed task replaces the approved store with a new database stack.
Use SQLAlchemy + PostgreSQL + Alembic -
04 DenyDENY
The pre-flight check matches two explicit anti-patterns and exits non-zero.
mneme_storage_json · add sqlalchemy · add migration layer · exit 2 -
05 Correction
The task is narrowed to add capability inside the existing storage boundary.
Add expires_at · keep JSON · no new dependencies -
06 AllowPASS
The same deterministic check runs clean on the corrected task.
Result: PASS · exit 0
Run it yourself
Run the violating and compliant tasks
Both inputs and the decision record ship with the product repository, so the deny-to-pass sequence is reproducible without a live model call.
git clone https://github.com/MnemeHQ/mneme
cd mneme
pip install -e .
mneme check --memory examples/project_memory.json \
--input examples/prompt_violation.txt --query "storage backend"
# FAIL [mneme_storage_json] · exit 2
mneme check --memory examples/project_memory.json \
--input examples/prompt_clean.txt --query "storage backend"
# Result: PASS · exit 0What happened
The decision names JSON as the approved store and ORM, SQLAlchemy, and migrations as anti-patterns.
The agent cannot create a parallel database, migration history, and credential path that later work would treat as established architecture.
The corrected task is not trusted automatically; it passes through the same decision and returns clean.
Common questions
Is the terminal output real?
Yes. The decision record and both input files are included in the Mneme repository. The transcript condenses the two real mneme check runs into one readable sequence.
Does Mneme rewrite the agent task?
No. Mneme blocks the violating task and identifies the decision it conflicts with. A human or the model submits the corrected task, which is evaluated through the same check.
Why block this before the agent writes code?
The proposal already contradicts an explicit architecture decision. Blocking it before execution prevents a predictable SQLAlchemy and migration diff from becoming review work.
Ready to try it?
Install Mneme and check your first decision in under five minutes.
pipx install "mneme-hq>=0.5.1"