{{PROJECT_NAME}} — Claude Code Instructions

Claude Code-specific configuration. Project conventions live in @AGENTS.md. Generated by SOVA (sova setup)

@AGENTS.md

Run Commands

Testing

# Run all tests
{{TEST_CMD}}

# Run a specific test
{{TEST_SPECIFIC_CMD}}

Linting and Formatting

# Lint
{{LINT_CMD}}

# Format
{{FORMAT_CMD}}

{{#DEV_CMD}}

Building / Running

# Start local environment
{{DEV_CMD}}

# Run migrations (if applicable)
{{MIGRATE_CMD}}

{{/DEV_CMD}}

{{#COMMON_COMMANDS}}

Other Commands

{{COMMON_COMMANDS}}

{{/COMMON_COMMANDS}}

Knowledge System (4-Tier)

Tier 1: Always-loaded context (no truncation)

• CLAUDE.md — This file (Claude Code-specific)
• AGENTS.md — Cross-cutting conventions for all AI tools
• .claude/rules/*.md — Modular knowledge files, loaded every session

Tier 1b: On-demand knowledge (loaded via commands)

• .claude/commands/*.md — Workflow commands and deep domain skills

Tier 2: Agent memory (persists across tasks, gitignored)

• .claude/agent-memory/MEMORY.md — Index file
• .claude/agent-memory/cookbook.md — Topical knowledge by domain, common mistakes with occurrence counts

Tier 3: Session memory (auto-managed by Claude Code)

• ~/.claude/projects/.../memory/MEMORY.md — User preferences, project state

Adding new knowledge

• Confirmed patterns: Add to .claude/rules/*.md (Tier 1)
• Deep domain knowledge: Create a command in .claude/commands/<topic>.md (Tier 1)
• New findings (not yet confirmed): Add to .claude/agent-memory/ (Tier 2)
• User preferences: Auto-memory handles this (Tier 3)
• Promote: Tier 2 → Tier 1 when confirmed in 2+ tasks

Agent Configuration

• Task source: {{TASK_SOURCE}}
• Model: {{AGENT_MODEL}}
• Max budget per task: ${{MAX_BUDGET}}
• Review rounds: {{REVIEW_MAX_ROUNDS}}

Agent Workflow

1. Read AGENTS.md and this file before starting any task
2. Understand the issue/task requirements fully before writing code
3. Follow TDD: write failing test → implement → refactor
4. Run all tests and linting before considering work done
5. Self-review your changes before pushing
6. Create a PR with a clear description linking to the issue
7. Record learnings in .claude/agent-memory/

Claude Code Behavioral Preferences

{{#NO_AI_COAUTHOR}}

• Do NOT include Co-Authored-By lines in commits
• Do NOT include "Generated with Claude Code" in PR descriptions {{/NO_AI_COAUTHOR}}
• Always run linter/formatter before creating commits
• Never push without explicit user approval {{EXTRA_PREFERENCES}}