Add database migrations for session context system

Phase 1: Database Schema Implementation
- Migration 010: Sessions table with bulletproof auto-incrementing
  - Unique session IDs across all projects
  - Per-project session numbers (1, 2, 3, ...)
  - Atomic sequence generation (no race conditions)
  - Session-task and session-commit linking
  - Semantic search with pgvector HNSW indexes

- Migration 011: Memories table with enhanced schema
  - Migrated existing session_memories to memories
  - Added session_id and task_id foreign keys
  - Renamed columns for consistency
  - HNSW indexing for semantic search

- Migration 012: Builds table for CI/CD tracking
  - Links builds to sessions and versions
  - Tracks build status, timing, and metadata

All migrations tested and verified on agiliton database.

Related: CF-167 (Fix shared session-summary.md file conflict)

migrations/010_sessions.sql

@@ -0,0 +1,170 @@
+-- Migration 010: Sessions table with bulletproof auto-incrementing
+-- Purpose: Store session metadata with unique session numbers per project
+-- Dependencies: 001_base_schema.sql (projects table, pgvector extension)
+
+-- Sessions table: Core session metadata with temporal tracking
+CREATE TABLE IF NOT EXISTS sessions (
+    id TEXT PRIMARY KEY,  -- Format: "session_{timestamp}_{uuid}"
+    project TEXT REFERENCES projects(key) ON DELETE SET NULL,
+    session_number INTEGER,  -- Auto-increment per project (e.g., "Session 439")
+
+    -- Temporal tracking
+    started_at TIMESTAMP WITH TIME ZONE NOT NULL,
+    ended_at TIMESTAMP WITH TIME ZONE,
+    duration_minutes INTEGER GENERATED ALWAYS AS
+        (EXTRACT(EPOCH FROM (ended_at - started_at)) / 60) STORED,
+
+    -- Context
+    working_directory TEXT,
+    git_branch TEXT,
+    initial_prompt TEXT,  -- First user message
+    summary TEXT,  -- Auto-generated summary
+
+    -- Semantic search
+    embedding vector(1024),  -- Embedding of summary for similarity search
+
+    -- Metrics
+    message_count INTEGER DEFAULT 0,
+    token_count INTEGER DEFAULT 0,
+    tools_used TEXT[],  -- Array of tool names used
+
+    -- Status
+    status TEXT DEFAULT 'active' CHECK (status IN ('active', 'completed', 'interrupted')),
+
+    created_at TIMESTAMP WITH TIME ZONE DEFAULT NOW(),
+    updated_at TIMESTAMP WITH TIME ZONE DEFAULT NOW()
+);
+
+-- Indexes for efficient querying
+CREATE INDEX idx_sessions_project ON sessions(project);
+CREATE INDEX idx_sessions_started ON sessions(started_at DESC);
+CREATE INDEX idx_sessions_status ON sessions(status);
+CREATE INDEX idx_sessions_number ON sessions(project, session_number DESC);
+
+-- HNSW index for semantic similarity search (requires pgvector)
+CREATE INDEX idx_sessions_embedding ON sessions USING hnsw (embedding vector_cosine_ops);
+
+-- Unique session number per project (partial index - only when project is set)
+CREATE UNIQUE INDEX idx_sessions_project_number ON sessions(project, session_number)
+    WHERE project IS NOT NULL;
+
+-- Session number sequences per project (bulletproof autoincrement)
+CREATE TABLE IF NOT EXISTS session_sequences (
+    project TEXT PRIMARY KEY REFERENCES projects(key) ON DELETE CASCADE,
+    next_number INTEGER DEFAULT 1,
+    last_updated TIMESTAMP WITH TIME ZONE DEFAULT NOW()
+);
+
+-- Function for atomic session number generation (prevents race conditions)
+CREATE OR REPLACE FUNCTION get_next_session_number(p_project TEXT)
+RETURNS INTEGER AS $$
+DECLARE
+    v_number INTEGER;
+BEGIN
+    -- Insert project if doesn't exist
+    INSERT INTO projects (key, name) VALUES (p_project, p_project)
+    ON CONFLICT (key) DO NOTHING;
+
+    -- Insert sequence if doesn't exist
+    INSERT INTO session_sequences (project, next_number)
+    VALUES (p_project, 1)
+    ON CONFLICT (project) DO NOTHING;
+
+    -- Atomically increment and return (no race conditions possible)
+    UPDATE session_sequences
+    SET next_number = next_number + 1,
+        last_updated = NOW()
+    WHERE project = p_project
+    RETURNING next_number - 1 INTO v_number;
+
+    RETURN v_number;
+END;
+$$ LANGUAGE plpgsql;
+
+-- Trigger to auto-assign session numbers on insert
+CREATE OR REPLACE FUNCTION assign_session_number()
+RETURNS TRIGGER AS $$
+BEGIN
+    IF NEW.project IS NOT NULL AND NEW.session_number IS NULL THEN
+        NEW.session_number := get_next_session_number(NEW.project);
+    END IF;
+    RETURN NEW;
+END;
+$$ LANGUAGE plpgsql;
+
+CREATE TRIGGER trg_assign_session_number
+    BEFORE INSERT ON sessions
+    FOR EACH ROW
+    EXECUTE FUNCTION assign_session_number();
+
+-- Session-Task relationship enhancement
+-- Add foreign key constraint to existing task_activity table
+ALTER TABLE task_activity
+    DROP CONSTRAINT IF EXISTS fk_task_activity_session;
+
+-- Drop NOT NULL constraint to allow NULL session_ids (for orphaned references)
+ALTER TABLE task_activity
+    ALTER COLUMN session_id DROP NOT NULL;
+
+-- First, set NULL for any session_ids that don't exist (orphaned references)
+-- This cleans up existing data before adding the constraint
+UPDATE task_activity
+SET session_id = NULL
+WHERE session_id IS NOT NULL
+  AND NOT EXISTS (SELECT 1 FROM sessions WHERE id = task_activity.session_id);
+
+ALTER TABLE task_activity
+    ADD CONSTRAINT fk_task_activity_session
+    FOREIGN KEY (session_id) REFERENCES sessions(id) ON DELETE SET NULL;
+
+-- Helper view: Session tasks with activity summary
+CREATE OR REPLACE VIEW session_tasks AS
+SELECT DISTINCT
+    ta.session_id,
+    ta.task_id,
+    t.project,
+    t.title,
+    t.status,
+    MIN(ta.created_at) as first_touched,
+    MAX(ta.created_at) as last_touched,
+    COUNT(*) as activity_count
+FROM task_activity ta
+JOIN tasks t ON ta.task_id = t.id
+GROUP BY ta.session_id, ta.task_id, t.project, t.title, t.status;
+
+-- Session-Commit linking table
+CREATE TABLE IF NOT EXISTS session_commits (
+    id SERIAL PRIMARY KEY,
+    session_id TEXT NOT NULL REFERENCES sessions(id) ON DELETE CASCADE,
+    commit_sha TEXT NOT NULL,
+    repo TEXT NOT NULL,
+    commit_message TEXT,
+    committed_at TIMESTAMP WITH TIME ZONE,
+    created_at TIMESTAMP WITH TIME ZONE DEFAULT NOW(),
+    UNIQUE(session_id, commit_sha)
+);
+
+CREATE INDEX idx_session_commits_session ON session_commits(session_id);
+CREATE INDEX idx_session_commits_sha ON session_commits(commit_sha);
+
+-- Helper view: Link sessions to versions through commits
+CREATE OR REPLACE VIEW session_versions AS
+SELECT DISTINCT
+    sc.session_id,
+    tc.task_id,
+    t.version_id,
+    v.version,
+    v.status as version_status
+FROM session_commits sc
+JOIN task_commits tc ON sc.commit_sha = tc.commit_sha
+JOIN tasks t ON tc.task_id = t.id
+LEFT JOIN versions v ON t.version_id = v.id
+WHERE v.id IS NOT NULL;
+
+-- Comments for documentation
+COMMENT ON TABLE sessions IS 'Session metadata with unique session numbers per project';
+COMMENT ON TABLE session_sequences IS 'Atomic counters for session numbers per project';
+COMMENT ON TABLE session_commits IS 'Links sessions to git commits';
+COMMENT ON COLUMN sessions.session_number IS 'Auto-incrementing number per project (1, 2, 3, ...)';
+COMMENT ON COLUMN sessions.embedding IS 'Vector embedding of session summary for semantic search';
+COMMENT ON FUNCTION get_next_session_number IS 'Atomic function to get next session number for a project';

migrations/011_memories.sql

@@ -0,0 +1,58 @@
+-- Migration 011: Memories/Learnings table with semantic search
+-- Purpose: Migrate existing session_memories to memories table with enhanced schema
+-- Dependencies: 001_base_schema.sql (pgvector), 010_sessions.sql (sessions table)
+
+-- Rename existing session_memories table to memories
+ALTER TABLE IF EXISTS session_memories RENAME TO memories;
+
+-- Add missing columns
+ALTER TABLE memories
+    ADD COLUMN IF NOT EXISTS task_id TEXT REFERENCES tasks(id) ON DELETE SET NULL;
+
+-- Rename source_session to session_id for consistency
+ALTER TABLE memories
+    RENAME COLUMN source_session TO session_id;
+
+-- Add foreign key constraint to sessions table
+ALTER TABLE memories
+    ADD CONSTRAINT fk_memories_session
+    FOREIGN KEY (session_id) REFERENCES sessions(id) ON DELETE SET NULL;
+
+-- Rename times_surfaced to access_count for consistency
+ALTER TABLE memories
+    RENAME COLUMN times_surfaced TO access_count;
+
+-- Rename last_surfaced to last_accessed_at for consistency
+ALTER TABLE memories
+    RENAME COLUMN last_surfaced TO last_accessed_at;
+
+-- Change varchar columns to TEXT for consistency
+ALTER TABLE memories
+    ALTER COLUMN project TYPE TEXT;
+
+-- Note: Keeping embedding as vector(1536) to preserve existing data
+-- Future embeddings can use vector(1024) by updating the column if needed
+
+-- Add missing indexes
+CREATE INDEX IF NOT EXISTS idx_memories_session ON memories(session_id);
+CREATE INDEX IF NOT EXISTS idx_memories_task ON memories(task_id);
+CREATE INDEX IF NOT EXISTS idx_memories_project ON memories(project);
+CREATE INDEX IF NOT EXISTS idx_memories_category ON memories(category);
+CREATE INDEX IF NOT EXISTS idx_memories_created ON memories(created_at DESC);
+
+-- Drop old index and create HNSW index for better performance
+-- Note: Existing index is ivfflat, we want hnsw
+DROP INDEX IF EXISTS idx_session_memories_embedding;
+CREATE INDEX IF NOT EXISTS idx_memories_embedding ON memories USING hnsw (embedding vector_cosine_ops);
+
+-- Full-text search on title and content (for keyword search)
+CREATE INDEX IF NOT EXISTS idx_memories_fts ON memories USING gin(
+    to_tsvector('english', title || ' ' || content)
+);
+
+-- Comments for documentation
+COMMENT ON TABLE memories IS 'Learnings and patterns discovered during development sessions';
+COMMENT ON COLUMN memories.category IS 'Type of memory: pattern (reusable solution), fix (bug resolution), preference (user choice), gotcha (trap/pitfall), architecture (design decision)';
+COMMENT ON COLUMN memories.embedding IS 'Vector embedding of title + content for semantic search';
+COMMENT ON COLUMN memories.context IS 'Optional context describing when/where this memory applies';
+COMMENT ON COLUMN memories.access_count IS 'Number of times this memory has been retrieved (for relevance ranking)';

migrations/012_builds.sql

@@ -0,0 +1,46 @@
+-- Migration 012: Builds table for CI/CD tracking
+-- Purpose: Track builds and link them to sessions and versions
+-- Dependencies: 001_base_schema.sql (versions table), 010_sessions.sql (sessions table)
+
+-- Builds table: Store build information linked to sessions and versions
+CREATE TABLE IF NOT EXISTS builds (
+    id SERIAL PRIMARY KEY,
+    session_id TEXT REFERENCES sessions(id) ON DELETE SET NULL,
+    version_id TEXT REFERENCES versions(id) ON DELETE CASCADE,
+
+    build_number INTEGER NOT NULL,
+    status TEXT DEFAULT 'pending' CHECK (status IN ('pending', 'running', 'success', 'failed')),
+
+    -- Build metadata
+    git_commit_sha TEXT,
+    git_branch TEXT,
+    build_log_url TEXT,
+    artifacts_url TEXT,
+
+    -- Timing
+    started_at TIMESTAMP WITH TIME ZONE NOT NULL,
+    finished_at TIMESTAMP WITH TIME ZONE,
+    duration_seconds INTEGER GENERATED ALWAYS AS
+        (EXTRACT(EPOCH FROM (finished_at - started_at))) STORED,
+
+    created_at TIMESTAMP WITH TIME ZONE DEFAULT NOW()
+);
+
+-- Indexes for efficient querying
+CREATE INDEX idx_builds_session ON builds(session_id);
+CREATE INDEX idx_builds_version ON builds(version_id);
+CREATE INDEX idx_builds_status ON builds(status);
+CREATE INDEX idx_builds_started ON builds(started_at DESC);
+CREATE INDEX idx_builds_commit ON builds(git_commit_sha);
+
+-- Unique constraint: one build number per version
+CREATE UNIQUE INDEX idx_builds_version_number ON builds(version_id, build_number)
+    WHERE version_id IS NOT NULL;
+
+-- Comments for documentation
+COMMENT ON TABLE builds IS 'CI/CD build tracking linked to sessions and versions';
+COMMENT ON COLUMN builds.session_id IS 'Optional link to session that triggered the build';
+COMMENT ON COLUMN builds.version_id IS 'Link to version being built';
+COMMENT ON COLUMN builds.duration_seconds IS 'Auto-calculated build duration in seconds';
+COMMENT ON COLUMN builds.build_log_url IS 'URL to build logs (e.g., GitHub Actions run)';
+COMMENT ON COLUMN builds.artifacts_url IS 'URL to build artifacts (e.g., app binary, Docker image)';