Limits & Constraints

System limits and constraints for Probe CLI and Probe Agent.

---

Probe CLI Limits

Search Limits

Limit	Default	Configuration
Max results	Unlimited	--max-results N
Max bytes	Unlimited	--max-bytes N
Max tokens	Unlimited	--max-tokens N
Timeout	30 seconds	--timeout N

Merge Threshold

Limit	Default	Configuration
Block merge distance	5 lines	--merge-threshold N

Session Caching

Limit	Default	Description
Cache size	Automatic	Based on available memory
Session lifetime	Until invalidation	Invalidated on file changes

---

Probe Agent Limits

Tool Iterations

Limit	Default	Configuration
Max iterations	30	maxIterations option

Timeouts

Limit	Default	Configuration
Request timeout	120,000 ms	requestTimeout option
Operation timeout	300,000 ms	maxOperationTimeout option
Engine activity	180,000 ms	ENGINE_ACTIVITY_TIMEOUT env

Engine Activity Timeout Range:

• Minimum: 5,000 ms (5 seconds)
• Maximum: 600,000 ms (10 minutes)
• Default: 180,000 ms (3 minutes)

Delegation Limits

Limit	Default	Configuration
Concurrent global	3	MAX_CONCURRENT_DELEGATIONS
Per session	10	MAX_DELEGATIONS_PER_SESSION
Queue timeout	60,000 ms	DELEGATION_QUEUE_TIMEOUT
Execution timeout	300 seconds	DELEGATION_TIMEOUT

Retry Limits

Limit	Default	Configuration
Max retries	3	retry.maxRetries
Initial delay	1,000 ms	retry.initialDelay
Max delay	30,000 ms	retry.maxDelay
Backoff factor	2	retry.backoffFactor

Retry Limits Range:

• Max retries: 0-100
• Initial delay: 0-60,000 ms
• Max delay: 0-300,000 ms
• Backoff factor: 1-10

Fallback Limits

Limit	Default	Configuration
Max total attempts	10	fallback.maxTotalAttempts

Range: 1-100

---

MCP Limits

Timeouts

Limit	Default	Configuration
Request timeout	30,000 ms	settings.timeout
Max timeout	1,800,000 ms	MCP_MAX_TIMEOUT

Max Timeout Range:

• Minimum: 30,000 ms (30 seconds)
• Maximum: 7,200,000 ms (2 hours)
• Default: 1,800,000 ms (30 minutes)

---

Provider Context Windows

Provider	Model	Context Window
Anthropic	Claude Sonnet	200,000 tokens
Anthropic	Claude Haiku	200,000 tokens
OpenAI	GPT-4o	128,000 tokens
OpenAI	GPT-4o-mini	128,000 tokens
Google	Gemini 1.5 Pro	1,000,000+ tokens
Google	Gemini 2.0 Flash	1,000,000+ tokens

---

Skill Limits

Limit	Value	Description
Skill name length	64 characters	Maximum skill name
Description length	400 characters	Auto-truncated
Skill name pattern	[a-z0-9-]+	Lowercase alphanumeric + hyphens

---

Task Limits

Limit	Description
Task ID format	task-1, task-2, ...
Dependencies	Must reference existing task IDs
Circular dependencies	Not allowed

---

Image Limits

Limit	Value	Description
Max file size	20 MB	Per image
Supported formats	PNG, JPG, JPEG, WebP, BMP, SVG

---

Bash Execution Limits

Probe Chat

Limit	Default	Configuration
Command timeout	120,000 ms	--bash-timeout

---

Memory Considerations

Large Codebases

For codebases with 100,000+ files:

1. Use result limits: --max-results 100
2. Use session pagination: --session my-search
3. Use language filters: --language rust
4. Increase Node.js memory: NODE_OPTIONS=--max-old-space-size=8192

Parser Pool

Setting	Default	Configuration
Pool size	CPU cores	PROBE_PARSER_POOL_SIZE
Tree cache	Automatic	PROBE_TREE_CACHE_SIZE

---

Rate Limiting

Provider Rate Limits

Rate limits are provider-specific. Common responses:

Error Code	Description	Solution
429	Rate limit exceeded	Enable retry with backoff
503	Service overloaded	Enable fallback to alternative

Retry Configuration

const agent = new ProbeAgent({
  retry: {
    maxRetries: 3,
    initialDelay: 1000,
    backoffFactor: 2
  }
});

---

Best Practices

1. Set Appropriate Limits

# For AI context windows
probe search "query" ./ --max-tokens 10000

# For large results
probe search "query" ./ --max-results 50 --session my-search

2. Use Pagination

# First batch
probe search "api" ./ --session api-search --max-results 100

# Subsequent batches
probe search "api" ./ --session api-search --max-results 100

3. Configure Timeouts

const agent = new ProbeAgent({
  requestTimeout: 120000,      // 2 minutes per request
  maxOperationTimeout: 600000  // 10 minutes total
});

4. Handle Resource Constraints

// Monitor token usage
const usage = agent.getTokenUsage();
if (usage.contextWindow > 150000) {
  await agent.compactHistory();
}

---

Related Documentation

• Performance Tuning - Optimization guide
• Environment Variables - Configuration
• Troubleshooting - Common issues