Skip to Content
CustomizationPer-rule export targeting

Per-rule export targeting

While AlignTrue syncs all your rules to all enabled exporters by default, you can target specific rules to specific agents using frontmatter options. This is useful for agent-specific guidance, debugging rules, or performance-critical configurations.

Why target rules per agent?

Common use cases:

  • Agent-specific guidance: Cursor-specific debugging strategies that don’t apply to other agents
  • Performance optimization: Memory or speed hints specific to one agent’s capabilities
  • Compliance requirements: Certain rules only needed for specific tools
  • Experimental features: Try new guidance in one agent before rolling out to all
  • Simplified exports: Keep exports minimal for single-agent projects by excluding irrelevant rules

How it works

Each rule file can specify which agents should receive it:

  • Use the exporter ids you have enabled in config (for example cursor, agents-md, windsurf, windsurf-mcp). Names must match exactly.
  • If you omit targeting, the rule goes to every enabled exporter.
---
title: Cursor Debugging Strategies
export_only_to: [cursor]
---
 
## Debugging in Cursor
 
Use Cursor's debugging features to trace execution...
---
title: General Architecture
exclude_from: []
---
 
## Architecture principles
 
These apply to all agents...

Frontmatter options

export_only_to

Explicitly list agents that should receive this rule. All other agents are excluded.

Syntax:

export_only_to: [cursor, claude]

Example:

---
title: TypeScript Performance Tips
export_only_to: [cursor, amazon-q, kilocode]
---
 
These agents support TypeScript with high performance...

Result: Only Cursor, Amazon Q, and KiloCode receive this rule. Copilot, Claude, etc. do not.

exclude_from

List agents that should NOT receive this rule. All other agents are included.

Syntax:

exclude_from: [cursor]

Example:

---
title: Generic AI Guidelines
exclude_from: [copilot]
---
 
General guidelines for all agents except Copilot...

Result: All agents receive this rule except Copilot.

Combining options

You can use both options together for clarity. The more restrictive setting wins:

---
title: Architecture Standards
export_only_to: [cursor, claude, amazon-q]
exclude_from: [claude]
---

In this case:

  • export_only_to says: only these 3 agents
  • exclude_from says: exclude claude
  • Result: Only Cursor and Amazon Q receive this rule

Common patterns

1. Shared core rules

Most rules apply to all agents:

---
title: Security Best Practices
---
 
Security guidelines for all agents...

No targeting needed. This goes everywhere.

2. Agent-specific advanced tips

For Cursor-specific features:

---
title: Cursor IDE Features
export_only_to: [cursor]
---
 
Leverage Cursor's built-in features...

3. Experimental or draft rules

Test new guidance in one agent:

---
title: Experimental Testing Approach
export_only_to: [cursor]
---
 
Testing approach currently under evaluation...

4. Legacy compatibility

Some rules needed for older agents:

---
title: Legacy Coding Standards
exclude_from: [claude]
---
 
For agents without modern language support...

5. Multi-agent teams with role separation

---
title: Frontend Guidelines
export_only_to: [cursor, copilot]
---
 
Frontend-specific guidance...

How targeting affects export

For native multi-file agents

Cursor, Amazon Q, KiloCode, Augment Code, Kiro, Trae AI:

  • If export_only_to: [cursor]: Only .cursor/rules/rule.mdc is written
  • If exclude_from: [cursor]: Rule is skipped, .cursor/rules/rule.mdc is not written
  • Otherwise: Rule writes to .cursor/rules/rule.mdc

Targets must match enabled exporters. If a rule only targets exporters that are not enabled, it is skipped and aligntrue sync logs a warning.

For AGENTS.md

Copilot, GitHub Copilot, Aider, Claude, etc.:

  • If export_only_to: [copilot]: Section appears in AGENTS.md
  • If exclude_from: [copilot]: Section does NOT appear in AGENTS.md
  • Otherwise: Section appears in AGENTS.md

Finding what each agent actually receives

Use the CLI to see which rules go to which agents:

aligntrue rules list --by-agent

Output:

Rules by agent:

  cursor (native format):
    - architecture.md
    - security.md
    - testing.md (via export_only_to)

  claude (native format):
    - architecture.md
    - security.md
    (excludes: testing.md via exclude_from)

  agents-md (Copilot, GitHub Copilot, etc.):
    - architecture.md
    - security.md

If you see a warning about rules targeting disabled exporters, enable the exporter in config or remove export_only_to from that rule.

Sync behavior

Targeting is evaluated on every sync:

  1. Edit a rule file: Change export_only_to or exclude_from
  2. Run sync: aligntrue sync
  3. Targeting updates: Agent files are regenerated based on new targeting
  4. Cleanup: If a rule is removed from an agent’s targeting, that agent’s file(s) are cleaned up

Example: Remove a rule from Cursor

Before:

# .aligntrue/rules/cursor-debug.md
---
title: Cursor Debugging
export_only_to: [cursor]
---

Exports to: .cursor/rules/cursor-debug.mdc

After (edit and sync):

# .aligntrue/rules/cursor-debug.md
---
title: Cursor Debugging
export_only_to: [claude] # Changed from cursor to claude
---

Next sync:

  • aligntrue sync regenerates exports
  • .cursor/rules/cursor-debug.mdc is removed (rule no longer targets Cursor)
  • Claude receives the rule instead

Best practices

When to use targeting

Use targeting when:

  • A rule is genuinely specific to one or a few agents
  • You want to minimize noise (don’t send irrelevant rules)
  • You’re testing new guidance in a subset of agents
  • Compliance or security requires limiting where rules go

Don’t use targeting for:

  • Normal rules that apply broadly (keep them universal)
  • Duplicating similar guidance per agent (consider a better approach)
  • Overly complex inclusion/exclusion logic (simplify instead)

Naming conventions

Make targeting obvious in rule filenames:

.aligntrue/rules/
├── architecture.md            # Universal, applies to all agents
├── cursor-specific.md         # Clearly Cursor-only
├── experimental-testing.md    # Experimental, check targeting in frontmatter
└── security.md                # Universal security

Documentation

Always document why targeting is needed:

---
title: Cursor Keyboard Shortcuts
export_only_to: [cursor]
description: Cursor-specific keyboard optimization (not applicable to other agents)
---
 
# Keyboard shortcuts in Cursor
 
Cursor has specific keyboard bindings...

Advanced: Combining with scopes

Targeting and scopes work together:

---
title: Frontend Testing
scope: apps/web # Only applies to apps/web
export_only_to: [cursor] # Only export to Cursor
---
 
Testing strategies for the web frontend...

Result: Only Cursor gets this rule, and only for the apps/web scope.

Troubleshooting

Rule not appearing for an agent

  1. Check export_only_to: Is the agent listed?
  2. Check exclude_from: Is the agent excluded?
  3. Check that the exporter is enabled in your config (run aligntrue rules list --by-agent to confirm names).
  4. Check enabled: Is the rule enabled? (defaults to true)
  5. Check scope: Does the scope match your project structure?
  6. Run aligntrue sync and look for warnings about skipped targets.

Rule appearing where it shouldn’t

  1. Verify export_only_to is set correctly
  2. Verify exclude_from doesn’t have typos
  3. Check agent/exporter names in targeting match config exactly
  4. Run aligntrue sync --verbose to see targeting decisions
Last updated on
OverlaysPlugs