Gives structure to AI edits
Zones define where agents may work, which dependencies matter, and what risk profile each part of the code has.
Airspace for modern software delivery
The codebase should decide what AI can touch.
gndctrl adds zone markers, stability tiers, and clearance rules directly into your project so agents can ship fast without breaking critical systems.
- 5 stability tiers
- 3 marker layers
- 1 persistent project memory
What gndctrl does
Enforces clearance, not just advice
Locked and sensitive zones require explicit handling, preventing accidental edits in high-impact surfaces.
Persists engineering memory
Node markers and CRID logbook entries retain non-obvious context between sessions so the next agent starts informed.
How it works
gndctrl uses a three-layer system: directory zones in `.gndctrl`, file-level `@gndctrl:zone` markers, and function-level `@gndctrl:node` markers for critical caveats. The result is enforceable context where it belongs: in and beside the code itself.
# @gndctrl:zone START | id=PAYMENT | stability=sensitive | deps=[AUTH_CORE]
# @gndctrl:node id=PAYMENT.charge_user | risk=high | crid=PMT-20260430-001
async def charge_user(...):
...
# @gndctrl:zone END | id=PAYMENTWhy this matters in today’s AI vibe-coded world
Velocity is no longer the bottleneck
AI can produce code instantly. The new bottleneck is trust: deciding whether that code should land in critical paths.
Context loss creates hidden risk
Most agent sessions are stateless. Without durable memory, every run repeats old mistakes and misses known edge cases.
Scaling teams need shared control planes
Single projects and platform fleets both need clear ownership, lock rules, and cross-zone coordination as AI usage grows.
gndctrl gives your codebase a voice
Instead of relying on prompts alone, governance becomes part of the repository contract and survives across tools and agents.
Ship fast. Keep control.
gndctrl is built for teams that want AI speed without surrendering architectural safety.
Start with zone markers