indusagi-coding-agent 0.1.28 → 0.1.30

package/docs/PRINT-MODE-GUIDE.md

@@ -1,456 +0,0 @@
-# Print Mode Complete Guide
-
-Comprehensive documentation for non-interactive mode used in scripting, CI/CD, and automation.
-
-## Table of Contents
-1. [Output Formats in Detail](#output-formats-in-detail)
-2. [Streaming Options](#streaming-options)
-3. [Error Reporting](#error-reporting)
-4. [Format Examples](#format-examples)
-5. [Performance Considerations](#performance-considerations)
-
----
-
-## Output Formats in Detail
-
-### JSON Mode (--mode=json)
-
-**Purpose:** Machine-readable event stream for integration with other tools.
-
-**Format:** NDJSON (Newline-Delimited JSON) - one JSON object per line
-
-**Events in Stream:**
-
-```json
-// Session initialization
-{"type": "session_created", "session_id": "abc-123", "timestamp": 1677000000000}
-
-// Prompting
-{"type": "message_start", "role": "user", "content": "Analyze this code"}
-{"type": "message_end", "role": "user"}
-
-// Agent processing
-{"type": "agent_start", "model": "gpt-4", "thinking_level": "medium"}
-{"type": "tool_start", "tool_name": "bash", "tool_input": {"command": "ls -la"}}
-{"type": "tool_output", "tool_name": "bash", "output": "total 48..."}
-{"type": "tool_end", "tool_name": "bash", "success": true}
-
-// Agent response
-{"type": "message_start", "role": "assistant", "content": "The code does..."}
-{"type": "message_end", "role": "assistant"}
-
-// Completion
-{"type": "agent_end", "total_tokens": 1523, "stop_reason": "end_turn"}
-```
-
-**Usage in Scripts:**
-
-```bash
-# Process all events
-indusagi --mode=json "prompt" | jq '.type'
-
-# Filter specific event types
-indusagi --mode=json "prompt" | jq 'select(.type == "tool_output")'
-
-# Extract final response
-indusagi --mode=json "prompt" | jq -s '.[] | select(.role == "assistant") | .content' -r
-
-# Calculate total tokens
-indusagi --mode=json "prompt" | jq -s '.[] | select(.type == "agent_end") | .total_tokens'
-```
-
-### TEXT Mode (--mode=text)
-
-**Purpose:** Human-readable formatted output for terminal display.
-
-**Format:**
-
-```
-┌─ Session: abc-123 ─────────────────────────────┐
-│ Model: gpt-4                                   │
-│ Mode: interactive                              │
-├────────────────────────────────────────────────┤
-│                                                │
-│ USER:                                          │
-│   Analyze this code                            │
-│                                                │
-│ AGENT:                                         │
-│   The code does...                             │
-│                                                │
-│ TOOLS EXECUTED:                                │
-│   - bash: ls -la (success)                     │
-│                                                │
-├────────────────────────────────────────────────┤
-│ Tokens: 1,523 (~1,142 words, ~$0.003 cost)   │
-│ Duration: 2.5s                                 │
-└────────────────────────────────────────────────┘
-```
-
-**Usage:**
-
-```bash
-# Display output
-indusagi --mode=text "What does this code do?"
-
-# Redirect to file
-indusagi --mode=text "prompt" > output.txt
-
-# Pipe to pager
-indusagi --mode=text "prompt" | less
-```
-
-### COMPACT Mode (--mode=compact)
-
-**Purpose:** Minimal output - just the final response text.
-
-**Format:** Plain text without formatting
-
-**Usage:**
-
-```bash
-# Get just the answer
-indusagi --mode=compact "What is 2+2?"
-# Output: 4
-
-# Use in scripting
-result=$(indusagi --mode=compact "Summarize: $file_content")
-echo "Summary: $result"
-
-# Chain multiple commands
-indusagi --mode=compact "List all Python files" | sort | uniq
-```
-
----
-
-## Streaming Options
-
-### Real-time Streaming (--stream)
-
-**Purpose:** Show output as it's generated (streaming mode).
-
-**Behavior:**
-- Displays tokens as they arrive from the model
-- Agent output updates in real-time
-- Tools show execution progress
-- Useful for long-running operations
-
-**Example:**
-
-```bash
-indusagi --mode=text --stream "Write a detailed essay on quantum computing"
-```
-
-**Output:** Displays essay line-by-line as generated
-
-### Buffered Output (default)
-
-**Behavior:**
-- Waits for agent to complete
-- Shows full response at once
-- Better for scripting (no partial output)
-- Lower CPU usage during execution
-
-**Example:**
-
-```bash
-indusagi --mode=json "prompt"
-```
-
-**Output:** All events printed once available
-
----
-
-## Error Reporting
-
-### Error Output to stderr
-
-All errors go to stderr while normal output goes to stdout:
-
-```bash
-# Capture errors separately
-indusagi --mode=json "prompt" 2>errors.log 1>output.jsonl
-
-# Display errors while piping output
-indusagi --mode=text "prompt" 2>&1 | tee output.txt
-```
-
-### Error JSON Format
-
-```json
-{
-  "type": "error",
-  "timestamp": 1677000000000,
-  "code": "MODEL_ERROR",
-  "message": "Rate limit exceeded",
-  "severity": "error",
-  "retriable": true,
-  "details": {
-    "model": "gpt-4",
-    "retry_after": 60
-  }
-}
-```
-
-### Error Types
-
-| Error Code | HTTP | Meaning | Retriable |
-|-----------|------|---------|-----------|
-| INVALID_ARGS | 400 | Invalid CLI arguments | No |
-| SESSION_NOT_FOUND | 404 | Session doesn't exist | No |
-| TOOL_ERROR | 500 | Tool execution failed | Yes |
-| MODEL_ERROR | 429/503 | Model quota/availability | Yes |
-| TIMEOUT | 504 | Operation exceeded timeout | Yes |
-| PERMISSION_DENIED | 403 | No access to resource | No |
-| INTERNAL_ERROR | 500 | Server error | Yes |
-
-### Error Handling in Scripts
-
-```bash
-#!/bin/bash
-
-# Capture both output and exit code
-output=$(indusagi --mode=json "prompt" 2>&1)
-exit_code=$?
-
-case $exit_code in
-  0)
-    echo "Success!"
-    echo "$output" | jq '.'
-    ;;
-  2)
-    echo "Invalid arguments"
-    echo "$output" | jq '.message'
-    ;;
-  3)
-    echo "Session not found"
-    ;;
-  4|5|6)
-    echo "Retriable error (code $exit_code), retrying..."
-    # Implement retry logic
-    ;;
-  *)
-    echo "Unknown error: $exit_code"
-    ;;
-esac
-```
-
----
-
-## Format Examples
-
-### Example 1: Code Summarization
-
-**Command:**
-```bash
-indusagi --mode=json @app.ts "Summarize this code" | jq -s '.[] | select(.role == "assistant") | .content' -r
-```
-
-**Output:**
-```
-The code implements a REST API server using Express.js with the following features:
-1. POST /api/users - Creates new users
-2. GET /api/users/:id - Retrieves user by ID
-3. DELETE /api/users/:id - Removes user
-...
-```
-
-### Example 2: CI/CD Integration
-
-**Script:**
-```bash
-#!/bin/bash
-# Check if code changes are breaking
-
-CHANGES=$(git diff HEAD~1 | head -1000)
-ANALYSIS=$(indusagi --mode=json <<EOF
-Review these code changes for breaking changes:
-$CHANGES
-EOF
-)
-
-if echo "$ANALYSIS" | jq -e '.[] | select(.content | contains("BREAKING"))' > /dev/null; then
-  echo "BREAKING CHANGES DETECTED"
-  exit 1
-else
-  echo "No breaking changes"
-  exit 0
-fi
-```
-
-### Example 3: Batch Processing
-
-**Script:**
-```bash
-#!/bin/bash
-# Process multiple files
-
-for file in *.ts; do
-  echo "Processing $file..."
-  
-  result=$(indusagi --mode=text --timeout=30 @"$file" "Generate unit tests")
-  
-  if [ $? -eq 0 ]; then
-    echo "$result" > "${file%.ts}.test.ts"
-    echo "✓ Generated tests"
-  else
-    echo "✗ Failed to generate tests"
-  fi
-done
-```
-
-### Example 4: Session Persistence
-
-**Command:**
-```bash
-# Create persistent session
-indusagi --session=my-project --mode=json "What is this codebase?"
-
-# Continue in same session
-indusagi --session=my-project --mode=json "How do I run tests?"
-
-# Export session
-indusagi --session=my-project --export=html > session.html
-```
-
----
-
-## Performance Considerations
-
-### Optimization Tips
-
-**1. Use Compact Mode for Simple Operations**
-```bash
-# Slower - parses full output
-result=$(indusagi --mode=text "Is this syntax valid?")
-
-# Faster - just the response
-result=$(indusagi --mode=compact "Is this syntax valid?")
-```
-
-**2. Batch Requests**
-```bash
-# Slower - multiple processes
-for item in ${items[@]}; do
-  indusagi --mode=compact "Process $item"
-done
-
-# Faster - single session
-indusagi --session=batch <<EOF
-Process item1
-Process item2
-Process item3
-EOF
-```
-
-**3. Use Session Caching**
-```bash
-# Without session - rebuilds context each time
-indusagi --mode=json "First question"
-indusagi --mode=json "Follow-up question"
-
-# With session - reuses context
-indusagi --session=work --mode=json "First question"
-indusagi --session=work --mode=json "Follow-up question"
-```
-
-**4. Limit Token Output**
-```bash
-# For large outputs, use jq to extract relevant parts
-indusagi --mode=json "Explain..." | jq '.[] | select(.role == "assistant") | .content[:500]'
-```
-
-### Memory Management
-
-**For Large File Processing:**
-```bash
-# Stream file instead of loading entirely
-head -c 100000 large_file.ts | indusagi --mode=compact "Analyze this code"
-
-# Or process in chunks
-split -b 50K large_file.ts chunk_
-for chunk in chunk_*; do
-  indusagi --mode=text @"$chunk" "What patterns do you see?"
-done
-```
-
-### Timeout Configuration
-
-**Default:** 5 minutes per request
-
-**Custom:**
-```bash
-indusagi --timeout=30 --mode=json "Quick question"
-indusagi --timeout=600 --mode=json "Complex analysis"
-```
-
-### Resource Limits
-
-**Recommended Defaults:**
-- Max input size: 100 KB
-- Max request timeout: 10 minutes
-- Max concurrent sessions: 10
-- Max output size: 5 MB
-
-**Configuration:**
-```json
-{
-  "print_mode": {
-    "max_input_size": 102400,
-    "timeout_ms": 600000,
-    "max_output_size": 5242880,
-    "streaming": false,
-    "format": "json"
-  }
-}
-```
-
----
-
-## Advanced Usage
-
-### Using jq for Output Processing
-
-```bash
-# Pretty print JSON
-indusagi --mode=json "..." | jq '.'
-
-# Extract all assistant messages
-indusagi --mode=json "..." | jq '.[] | select(.role == "assistant") | .content'
-
-# Calculate statistics
-indusagi --mode=json "..." | jq -s '[.[] | select(.type == "agent_end") | .total_tokens] | add'
-
-# Format as table
-indusagi --mode=json "..." | jq -r '.[].content' | column -t -s '|'
-```
-
-### Combining with Other Tools
-
-```bash
-# Parse with grep
-indusagi --mode=json "..." | jq -r '.content' | grep "ERROR"
-
-# Process with sed
-indusagi --mode=text "..." | sed 's/old/new/g'
-
-# Sort and unique
-indusagi --mode=compact "..." | sort | uniq
-
-# Count lines
-indusagi --mode=text "..." | wc -l
-```
-
-### Shell Integration
-
-```bash
-# Save common commands
-alias analyze='indusagi --mode=compact --timeout=30'
-
-# Create wrapper function
-function code_review() {
-  indusagi --session=review --mode=json "@$1" "Code review: suggest improvements"
-}
-
-# Use in scripts
-code_review ./main.ts | jq '.[] | select(.content | contains("TODO"))'
-```

package/docs/README.md

@@ -1,78 +0,0 @@
-# Indusagi Coding Agent
-
-A terminal-first coding agent with a small core and strong extensibility. Use it interactively, in scripts (print/JSON), over RPC, or as an SDK in your own apps.
-
-## Install
-
-```bash
-npm install -g indusagi-coding-agent
-```
-
-## Quick Start
-
-```bash
-indusagi
-```
-
-Authenticate using `/login` or set provider API keys.
-
-## Modes
-
-| Mode | Description | Docs |
-|------|-------------|------|
-| Interactive | Default mode with TUI | - |
-| Print/JSON | For scripting | [json.md](json.md) |
-| RPC | Process integration | [rpc.md](rpc.md) |
-| SDK | Programmatic usage | [sdk.md](sdk.md) |
-
-## Key Features
-
-- **Multi-provider support** - Anthropic, OpenAI, Google, and more
-- **Extensible** - Extensions, skills, hooks, and themes
-- **Session management** - Tree-based navigation, branching, compaction
-- **Tool system** - Built-in tools for files, bash, search
-- **Subagents** - Spawn specialized agents for complex tasks
-
-## Documentation Map
-
-### Getting Started
-- [Providers](providers.md) - Configure LLM providers
-- [Settings](settings.md) - Global and project settings
-
-### SDK & API
-- [SDK Reference](sdk.md) - Programmatic usage
-- [RPC Mode](rpc.md) - JSON-RPC integration
-- [JSON Mode](json.md) - Scripting output
-
-### Customization
-- [Extensions](extensions.md) - Extend agent behavior
-- [Skills](skills.md) - On-demand capability packages
-- [Hooks](hooks.md) - Low-level interception
-- [Subagents](subagents.md) - Spawn specialized agents
-- [Prompt Templates](prompt-templates.md) - Custom prompts
-- [Themes](themes.md) - UI theming
-- [Packages](packages.md) - Share resources
-
-### Session Management
-- [Session Format](session.md) - File format and structure
-- [Tree Navigation](tree.md) - Branch and navigate history
-- [Compaction](compaction.md) - Context management
-
-### Configuration
-- [Settings](settings.md) - All configuration options
-- [Custom Models](models.md) - Add custom providers
-- [Custom Providers](custom-provider.md) - Proxy support
-- [Keybindings](keybindings.md) - Keyboard shortcuts
-
-### UI
-- [TUI Components](tui.md) - Terminal UI
-- [Terminal Setup](terminal-setup.md) - Terminal configuration
-
-### Platform
-- [Development](development.md) - Contributing
-- [Shell Aliases](shell-aliases.md) - Shell integration
-- [Windows](windows.md) - Windows-specific notes
-
-## License
-
-MIT