qc-validator-mcp
# qc-validator-mcp
Runtime quality validation for AI agent outputs. Detect hallucinations, enforce scope compliance, and score output quality — all via MCP.
## Install
```bash
npx qc-validator-mcp
```
### Claude Desktop
```json
{
"mcpServers": {
"qc-validator": {
"command": "npx",
"args": ["qc-validator-mcp"]
}
}
}
```
## Tools
### validate_output
Score agent output against configurable criteria: length limits, required keywords, forbidden patterns, and factual claim density.
```
Params: output, task_description, criteria { max_length, required_keywords[], forbidden_patterns[], factual_claims_count }
Returns: { pass, score, issues[], recommendation }
```
### check_hallucination_risk
Estimate hallucination likelihood. With source text, checks sentence-level grounding. Without source, flags outputs dense with specific numbers, dates, and URLs.
```
Params: output, source_text (optional), claim_count (default 5)
Returns: { risk_level, unsupported_claims[], confidence, suggestion }
```
### check_scope_compliance
Validate output against a scope contract — allowed/forbidden topics, word limits, required sections.
```
Params: output, scope { allowed_topics[], forbidden_topics[], max_words, required_sections[] }
Returns: { compliant, violations[], scope_utilization_percent }
```
### log_validation
Store validation results for per-agent trending.
```
Params: agent_id, output_hash, score, pass, issues_count
Returns: { logged, agent_id, total_validations }
```
### get_failure_patterns
Analyze common failure modes for a specific agent.
```
Params: agent_id
Returns: { total_validations, pass_rate, avg_score, most_common_issues[], trend }
```
### generate_quality_report
Quality dashboard across all validated agents — no parameters required.
```
Returns: { total_agents, overall_pass_rate, agents[], worst_performers[], best_performers[], recommendations[] }
```
## Resource
- `qc://dashboard` — Quality metrics for all validated agents
## Architecture
- Pure Node.js ES modules
- In-memory Maps (no external dependencies)
- stdio transport via @modelcontextprotocol/sdk
- Zero configuration required
## License
MIT
TDQS
Scored across 6 tools
The tools are mostly distinct: validate_output provides general scoring, while check_hallucination_risk and check_scope_compliance target specific failure modes. There is slight overlap between validate_output's forbidden patterns and check_scope_compliance's allowed/forbidden topics, but the descriptions clarify different scopes.
All tool names follow a consistent verb_noun pattern (validate_output, get_failure_patterns, check_hallucination_risk, check_scope_compliance, log_validation, generate_quality_report), using lowercase with underscores throughout. This makes the API predictable and easy to navigate.
With 6 tools, the server is well-scoped for its purpose of validating agent output. Each tool covers a distinct step in the validation workflow: executing checks, analyzing patterns, logging results, and reporting, without unnecessary bloat.
The core workflow (validate, log, analyze, report) is covered. However, there is no tool to manage the configurable criteria mentioned in validate_output, which is a minor gap. Additionally, update/delete operations for logged results are missing, but the main lifecycle is intact.