Agent Alignment Layer

The Agent Alignment Layer lets AI coding assistants (Claude Code, GitHub Copilot, Pylon, etc.) check whether a proposed action conflicts with your team's documented decisions BEFORE taking it autonomously.

How It Works

1. Agent proposes an action (create Jira ticket, open PR, post Slack message)
2. Agent calls align.check_proposed_action or POST /alignment/check
3. Align searches your decision graph using semantic similarity (fast, ~300ms)
4. If a potential conflict is found, an LLM confirms and explains it
5. Agent receives: aligned / conflicting / no-context + specific conflict details

MCP Tool: align.check_proposed_action

For AI assistants connected to the Align MCP server:

Input:

{
  "action_type": "jira_ticket",
  "content": "Migrate user service to MongoDB for flexible schema",
  "context": "Team: Backend, Project: AUTH"
}

Response (conflicting):

{
  "status": "conflicting",
  "confidence": 0.91,
  "relevant_decisions": [
    {
      "id": "d-42",
      "title": "Use PostgreSQL for all services",
      "summary": "Team decided Postgres is our only data store",
      "status": "active",
      "similarity": 0.87
    }
  ],
  "conflicts": [
    {
      "decision_id": "d-42",
      "title": "Use PostgreSQL for all services",
      "reason": "Proposes MongoDB which contradicts the Postgres-only decision",
      "severity": "critical",
      "suggested_resolution": "Review decision d-42 before proceeding"
    }
  ],
  "message": "This action conflicts with 1 existing decision(s). Review the conflicts before proceeding."
}

REST API: POST /alignment/check

For CI pipelines, custom bots, and direct integrations:

curl -X POST https://api.align.tech/alignment/check \
  -H "Authorization: Bearer <YOUR_API_KEY>" \
  -H "x-tenant-id: <YOUR_TENANT_ID>" \
  -H "Content-Type: application/json" \
  -d '{
    "action_type": "pull_request",
    "content": "Add MongoDB driver to user service",
    "context": "Relates to JIRA-1234"
  }'

Response Shape

Field	Type	Description
status	aligned | conflicting | no-context	Alignment result
confidence	0-1	Confidence in the result
relevant_decisions	array	Related decisions found
conflicts	array?	Present only when status=conflicting
message	string	Human-readable summary for agents

Status Meanings

Status	Meaning	Agent should...
aligned	Action aligns with or is unrelated to existing decisions	Proceed normally
conflicting	Action contradicts one or more existing decisions	Surface conflicts to user, request confirmation
no-context	No relevant decisions found	Proceed - consider documenting the decision afterward

Action Types

action_type	Use for
jira_ticket	Jira issue creation
pull_request	GitHub/GitLab PR description
slack_message	Slack or Teams messages
commit_message	Git commit messages
general	Any other action

Performance

• Fast path (vector search only): ~300-500ms - returned when no strong match found
• LLM path (conflict confirmed): ~1.5-2.5s - returns specific conflict details
• Hard LLM timeout: 1.5s - falls back to aligned with confidence: 0.5 if exceeded

Agent Integration Example

When an agent is about to create a Jira ticket:

const check = await mcp.callTool('align.check_proposed_action', {
  action_type: 'jira_ticket',
  content: `${ticket.title}\n\n${ticket.description}`,
  context: `Project: ${ticket.project}`,
});

if (check.status === 'conflicting') {
  return `This ticket conflicts with existing decisions:\n${
    check.conflicts.map(c => `- ${c.title}: ${c.reason}`).join('\n')
  }\n\nProceed anyway?`;
}

Setup

The align.check_proposed_action tool is available on the Align MCP server (port 8089). Connect your AI assistant to the Align MCP server - see the AI Assistants (MCP) guide for setup instructions.

For direct REST API access, authenticate with your API key and tenant ID as shown in the REST example above.