Context Nexus
What we want Context Injection System Prompt
Intent register
Specs, Decision Directives, roadmap, and product hypotheses. The precise instructions AI agents execute to generate code, tests, and components.
What does Intent contain?
Specs & user stories · Acceptance criteria · Mockups · Decision Directives · OKRs & roadmap · Personas · Product hypotheses.
The fundamental invariant
Humans vs AI Agents
Human
Formulates intent, aligns the team on what is being built, prioritizes, decides trade-offs.
AI Agent
Consumes as direct instruction to generate code, tests, components, migrations. The precision of this register directly determines output quality.
Asymmetry: formulation vs. execution.
Consumption modes
System Prompt
Universal Decision Directives (applicable to all tickets) go in the system prompt.
Context Injection
The current spec, user stories, and task-specific Decision Directives are passed by context injection.
Decision Directive structure
intent/directives/event-driven-indexation.md YAML + Markdown
---
register: intent
level: team
owner: tech-lead
status: active
consumption-mode: system-prompt
last-validated: 2026-06-03
supersedes: null
exception-to: null
---
# Decision Directive — Event-driven indexation
All hotel catalogue updates must transit through a Kafka event
before any write to Elasticsearch. No synchronous catalogue write
to Elasticsearch is allowed in this bounded context.
Acceptance criteria: see Contracts/search-indexation-sla.md
Decision context: Knowledge/decision-contexts/event-driven-indexation.mdSpec structure in Intent
intent/specs/SEARCH-89.md YAML + Markdown
---
register: intent
consumption-mode: context-injection
---
# Spec SEARCH-89
## Objective
Reduce P95 latency of the results page to under 200ms.
## Acceptance criteria
- [ ] P95 measured on the indexation pipeline: ≤ 200ms
- [ ] No synchronous Elasticsearch write (see Decision Directive /directives/event-driven)
- [ ] Load test: 500 req/s for 5 minutes without degradation See in practice