diff --git a/.agents/sessions/2026-02-27-session-1.json b/.agents/sessions/2026-02-27-session-1.json
new file mode 100755
index 000000000..51f60d808
--- /dev/null
+++ b/.agents/sessions/2026-02-27-session-1.json
@@ -0,0 +1,132 @@
+{
+  "session": {
+    "number": 1,
+    "date": "2026-02-27",
+    "branch": "feat/592-autonomous",
+    "startingCommit": "79e3b5f",
+    "objective": "AO-002: Define Agent Registry Schema - Create TypeScript interfaces for agent registry per ADR-013"
+  },
+  "protocolCompliance": {
+    "sessionStart": {
+      "serenaActivated": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "Serena activated successfully"
+      },
+      "serenaInstructions": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "Serena instructions loaded"
+      },
+      "handoffRead": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "HANDOFF.md content reviewed"
+      },
+      "sessionLogCreated": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "This file"
+      },
+      "skillScriptsListed": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "Skill scripts enumerated"
+      },
+      "usageMandatoryRead": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "Usage mandatory content loaded"
+      },
+      "constraintsRead": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "PROJECT-CONSTRAINTS.md reviewed"
+      },
+      "memoriesLoaded": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "Relevant memories retrieved"
+      },
+      "branchVerified": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "feat/592-autonomous"
+      },
+      "notOnMain": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "On feat/592-autonomous"
+      },
+      "gitStatusVerified": {
+        "level": "SHOULD",
+        "Complete": false,
+        "Evidence": ""
+      },
+      "startingCommitNoted": {
+        "level": "SHOULD",
+        "Complete": true,
+        "Evidence": "79e3b5f"
+      }
+    },
+    "sessionEnd": {
+      "checklistComplete": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "All MUST items marked complete"
+      },
+      "handoffNotUpdated": {
+        "level": "MUST NOT",
+        "Complete": false,
+        "Evidence": "HANDOFF.md not modified"
+      },
+      "serenaMemoryUpdated": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "Serena memory updated with session learnings"
+      },
+      "markdownLintRun": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "No markdown files modified"
+      },
+      "changesCommitted": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "Commit c48d6d3"
+      },
+      "validationPassed": {
+        "level": "MUST",
+        "Complete": true,
+        "Evidence": "Validation script exit code 0"
+      },
+      "tasksUpdated": {
+        "level": "SHOULD",
+        "Complete": false,
+        "Evidence": ""
+      },
+      "retrospectiveInvoked": {
+        "level": "SHOULD",
+        "Complete": false,
+        "Evidence": ""
+      }
+    }
+  },
+  "workLog": [
+    {
+      "timestamp": "2026-02-27T05:19:00Z",
+      "action": "Created TypeScript schema for agent registry per ADR-013",
+      "outcome": "Schema file created with 22 agents"
+    },
+    {
+      "timestamp": "2026-02-27T06:00:00Z",
+      "action": "PR monitor: Fixed AgentName to match AGENTS.md canonical catalog",
+      "outcome": "Removed debug/janitor/spec-generator, added backlog-generator (20 agents)"
+    }
+  ],
+  "endingCommit": "pending",
+  "nextSteps": [
+    "Commit fixes and push",
+    "Verify CI passes"
+  ]
+}
\ No newline at end of file
diff --git a/src/agent-registry-schema.ts b/src/agent-registry-schema.ts
new file mode 100644
index 000000000..e6745a446
--- /dev/null
+++ b/src/agent-registry-schema.ts
@@ -0,0 +1,495 @@
+/**
+ * Agent Registry Schema
+ *
+ * TypeScript interfaces for agent registry per ADR-013.
+ * This defines the structure for agent definitions, invocation parameters,
+ * and orchestration primitives.
+ */
+
+/**
+ * Valid agent names in the system.
+ * Canonical source: AGENTS.md Agent Catalog table.
+ */
+export type AgentName =
+  | "analyst"
+  | "architect"
+  | "backlog-generator"
+  | "critic"
+  | "devops"
+  | "explainer"
+  | "high-level-advisor"
+  | "implementer"
+  | "independent-thinker"
+  | "memory"
+  | "merge-resolver"
+  | "milestone-planner"
+  | "orchestrator"
+  | "pr-comment-responder"
+  | "qa"
+  | "retrospective"
+  | "roadmap"
+  | "security"
+  | "skillbook"
+  | "task-decomposer";
+
+/**
+ * Available Claude model types for agent execution.
+ */
+export type ModelName = "sonnet" | "opus" | "haiku";
+
+/**
+ * Input specification for an agent.
+ */
+export interface InputSpec {
+  name: string;
+  type: string;
+  description: string;
+  required: boolean;
+}
+
+/**
+ * Output specification for an agent.
+ */
+export interface OutputSpec {
+  name: string;
+  type: string;
+  description: string;
+}
+
+/**
+ * Complete agent definition including capabilities and relationships.
+ */
+export interface AgentDefinition {
+  /** Agent name (must match AgentName type) */
+  name: AgentName;
+
+  /** Path to agent definition file */
+  file: string;
+
+  /** One-line role description */
+  role: string;
+
+  /** Domain expertise and specialization */
+  specialization: string;
+
+  /** Default Claude model for this agent */
+  default_model: ModelName;
+
+  /** Input specifications */
+  inputs: InputSpec[];
+
+  /** Output specifications */
+  outputs: OutputSpec[];
+
+  /** Agents this agent can delegate work to */
+  delegates_to: AgentName[];
+
+  /** Agents that can invoke this agent */
+  called_by: AgentName[];
+
+  /** Optional artifact output directory */
+  artifact_directory?: string;
+}
+
+/**
+ * Context passed between agents during handoffs.
+ */
+export interface AgentContext {
+  /** Agent that performed the handoff */
+  prior_agent?: AgentName;
+
+  /** Summary of prior agent's output */
+  prior_output?: string;
+
+  /** Paths to relevant artifacts */
+  artifacts?: string[];
+
+  /** Injected steering files */
+  steering?: string[];
+
+  /** Link to Session State MCP session ID */
+  session_id?: string;
+}
+
+/**
+ * Parameters for invoking an agent.
+ */
+export interface InvokeAgentParams {
+  /** Agent to invoke */
+  agent: AgentName;
+
+  /** Task prompt for the agent */
+  prompt: string;
+
+  /** Additional context for the agent */
+  context?: AgentContext;
+
+  /** Override default model */
+  model_override?: ModelName;
+
+  /** Parallel execution tracking ID */
+  parallel_id?: string;
+}
+
+/**
+ * Result from agent invocation.
+ */
+export interface InvokeAgentResult {
+  /** Unique invocation reference */
+  invocation_id: string;
+
+  /** Agent that was invoked */
+  agent: AgentName;
+
+  /** Model used for execution */
+  model: ModelName;
+
+  /** Invocation status */
+  status: "started" | "completed" | "failed";
+
+  /** Agent output (if completed) */
+  output?: string;
+
+  /** Artifacts created by agent */
+  artifacts_created?: string[];
+
+  /** Recommended next agents */
+  suggested_next?: AgentName[];
+
+  /** Context for next agent handoff */
+  handoff_context?: HandoffContext;
+}
+
+/**
+ * Decision record with rationale.
+ */
+export interface Decision {
+  /** Decision made */
+  decision: string;
+
+  /** Rationale for the decision */
+  rationale: string;
+
+  /** Alternatives that were considered */
+  alternatives_considered?: string[];
+}
+
+/**
+ * Context for agent-to-agent handoff.
+ */
+export interface HandoffContext {
+  /** Summary of what was accomplished */
+  summary: string;
+
+  /** Artifacts created */
+  artifacts: string[];
+
+  /** Key decisions made */
+  decisions: Decision[];
+
+  /** Unresolved questions */
+  open_questions: string[];
+
+  /** Recommendations for next agent */
+  recommendations: string[];
+}
+
+/**
+ * Parameters for tracking agent handoff.
+ */
+export interface TrackHandoffParams {
+  /** Agent performing handoff */
+  from_agent: AgentName;
+
+  /** Agent receiving handoff */
+  to_agent: AgentName;
+
+  /** Handoff context */
+  context: HandoffContext;
+
+  /** Parallel execution tracking ID */
+  parallel_id?: string;
+}
+
+/**
+ * Result from tracking handoff.
+ */
+export interface TrackHandoffResult {
+  /** Unique handoff reference */
+  handoff_id: string;
+
+  /** Handoff timestamp */
+  timestamp: string;
+
+  /** Source agent */
+  from: AgentName;
+
+  /** Destination agent */
+  to: AgentName;
+
+  /** Whether context was preserved */
+  context_preserved: boolean;
+
+  /** Session ID (if applicable) */
+  session_id?: string;
+}
+
+/**
+ * Workflow definition.
+ */
+export interface WorkflowDefinition {
+  /** Workflow name */
+  name: string;
+
+  /** When to use this workflow */
+  trigger: string;
+
+  /** Ordered agent sequence */
+  agents: AgentName[];
+
+  /** Quality gates in workflow */
+  gates: string[];
+}
+
+/**
+ * Routing rule for agent selection.
+ */
+export interface RoutingRule {
+  /** Pattern to match in task description (regex string) */
+  pattern: string;
+
+  /** Primary agent for this pattern */
+  primary: AgentName;
+
+  /** Fallback agent if primary unavailable */
+  fallback: AgentName;
+
+  /** Confidence level (0-1) */
+  confidence: number;
+}
+
+/**
+ * Agent catalog result.
+ */
+export interface GetAgentCatalogResult {
+  /** All agent definitions */
+  agents: AgentDefinition[];
+
+  /** Available workflows */
+  workflows: WorkflowDefinition[];
+
+  /** Routing heuristics */
+  routing_heuristics: RoutingRule[];
+}
+
+/**
+ * Alternative workflow option.
+ */
+export interface Alternative {
+  /** Workflow name */
+  workflow: string;
+
+  /** Agent sequence */
+  agents: AgentName[];
+
+  /** Confidence level (0-100) */
+  confidence: number;
+
+  /** Reason not chosen */
+  why_not_chosen: string;
+}
+
+/**
+ * Parameters for routing recommendation.
+ */
+export interface GetRoutingRecommendationParams {
+  /** Task description */
+  task: string;
+
+  /** Current session state */
+  current_state?: string;
+
+  /** Task constraints */
+  constraints?: string[];
+}
+
+/**
+ * Routing recommendation result.
+ */
+export interface GetRoutingRecommendationResult {
+  /** Recommended workflow name */
+  recommended_workflow: string;
+
+  /** Recommended agent sequence */
+  recommended_agents: AgentName[];
+
+  /** Confidence level (0-100) */
+  confidence: number;
+
+  /** Reasoning for recommendation */
+  reasoning: string;
+
+  /** Alternative options */
+  alternatives: Alternative[];
+}
+
+/**
+ * Agent for parallel execution.
+ */
+export interface ParallelAgent {
+  /** Agent to invoke */
+  agent: AgentName;
+
+  /** Task prompt */
+  prompt: string;
+
+  /** Additional context */
+  context?: AgentContext;
+}
+
+/**
+ * Parameters for parallel execution.
+ */
+export interface StartParallelExecutionParams {
+  /** Agents to execute in parallel */
+  agents: ParallelAgent[];
+
+  /** Result aggregation strategy */
+  aggregation_strategy: "merge" | "vote" | "escalate";
+
+  /** Agent that resolves conflicts */
+  conflict_handler?: AgentName;
+}
+
+/**
+ * Parallel execution result.
+ */
+export interface StartParallelExecutionResult {
+  /** Parallel execution ID */
+  parallel_id: string;
+
+  /** Agents that were started */
+  agents_started: AgentName[];
+
+  /** Execution status */
+  status: "running" | "completed" | "conflict";
+
+  /** Estimated completion timestamp */
+  estimated_completion?: string;
+}
+
+/**
+ * Result from individual parallel agent.
+ */
+export interface ParallelResult {
+  /** Agent that executed */
+  agent: AgentName;
+
+  /** Execution status */
+  status: "completed" | "failed" | "pending";
+
+  /** Agent output */
+  output?: string;
+
+  /** Artifacts created */
+  artifacts?: string[];
+}
+
+/**
+ * Position in a conflict.
+ */
+export interface Position {
+  /** Agent holding this position */
+  agent: AgentName;
+
+  /** Position statement */
+  position: string;
+
+  /** Supporting evidence */
+  evidence?: string[];
+}
+
+/**
+ * Conflict between agents.
+ */
+export interface Conflict {
+  /** Agents in conflict */
+  agents: AgentName[];
+
+  /** Issue description */
+  issue: string;
+
+  /** Positions held by agents */
+  positions: Position[];
+}
+
+/**
+ * Parameters for aggregating parallel results.
+ */
+export interface AggregateParallelResultsParams {
+  /** Parallel execution ID */
+  parallel_id: string;
+
+  /** Wait for all agents to complete */
+  wait_for_all?: boolean;
+}
+
+/**
+ * Aggregated parallel results.
+ */
+export interface AggregateParallelResultsResult {
+  /** Parallel execution ID */
+  parallel_id: string;
+
+  /** Aggregation status */
+  status: "complete" | "partial" | "conflict";
+
+  /** Individual agent results */
+  results: ParallelResult[];
+
+  /** Merged output */
+  aggregated_output?: string;
+
+  /** Conflicts detected */
+  conflicts?: Conflict[];
+
+  /** Whether resolution is needed */
+  resolution_needed?: boolean;
+}
+
+/**
+ * Parameters for conflict resolution.
+ */
+export interface ResolveConflictParams {
+  /** Parallel execution ID */
+  parallel_id: string;
+
+  /** Conflict ID to resolve */
+  conflict_id: string;
+
+  /** Resolution strategy */
+  resolution_strategy: "vote" | "escalate" | "manual";
+
+  /** Manual resolution (if strategy is manual) */
+  manual_resolution?: string;
+}
+
+/**
+ * Conflict resolution result.
+ */
+export interface ResolveConflictResult {
+  /** Conflict ID */
+  conflict_id: string;
+
+  /** Whether resolved */
+  resolved: boolean;
+
+  /** Resolution details */
+  resolution: string;
+
+  /** Who/what resolved it */
+  resolved_by: AgentName | "manual";
+
+  /** Rationale for resolution */
+  rationale: string;
+}