Answers from the documentation; checking which model is on duty
Hi - I answer from the OpenSmartRoute documentation: routing, the API, plans and quotas, self-hosting. Ask away, or open a support ticket if you need a person.
Grounded in the docs - follow a source before acting on it.
qe-pentest-validation - Skill - OpenSmartRoute
Skillv1.0.0
qe-pentest-validation
Orchestrate security finding validation through graduated exploitation. 4-phase pipeline: recon (SAST/DAST), analysis (code review), validation (exploit proof), report (No Exploit, No Report gate).
Imported from proffesor-for-testing/agentic-qe (.kiro/skills/qe-pentest-validation/SKILL.md). Install upstream with npx skills add proffesor-for-testing/agentic-qe --skill qe-pentest-validation. Copyright stays with the author.
Pentest Validation
<default_to_action>
When validating security findings:
REQUIRE explicit authorization for target URL
SCAN with qe-security-scanner (SAST + dependency + secrets)
ANALYZE with qe-security-reviewer + qe-security-auditor (parallel)
VALIDATE with qe-pentest-validator (graduated exploitation, parallel per vuln type)
REPORT only confirmed findings with PoC evidence ("No Exploit, No Report")
UPDATE exploit playbook with new patterns
Quality Gates:
Authorization confirmed before ANY exploitation
Target URL is staging/dev (NOT production)
Budget cap enforced ($15 default)
Time cap enforced (30 min default)
All exploitation attempts logged
</default_to_action>
Quick Reference Card
The 4-Phase Pipeline
Phase
Agent(s)
Purpose
Parallelism
1. Recon
qe-security-scanner
SAST, DAST, dependency scan, secrets
Internal parallel
2. Analysis
qe-security-reviewer + qe-security-auditor
Code review + compliance check
Both in parallel
3. Validation
qe-pentest-validator
Graduated exploit validation
Per-vuln-type parallel
4. Report
qe-quality-gate
"No Exploit, No Report" filter
Sequential
Graduated Exploitation Tiers
Tier
Handler
Cost
Latency
Use When
1
Agent Booster (WASM)
$0
<1ms
Code pattern is conclusive (eval, innerHTML, hardcoded creds)
2
Haiku
$0.0002
~500ms
Need payload test against live target
3
Sonnet/Opus
$0.003-$0.015
2-5s
Full exploit chain with data proof
When to Use This Skill
Scenario
Tier
Estimated Cost
PR security review (source only)
1
$0
Pre-release validation (staging)
1-2
$1-5
Full pentest validation
1-3
$5-15
Compliance audit evidence
1-3
$5-15
Configuration
pentest: target_url: https://staging.app.com # REQUIRED for Tier 2-3 source_repo: ./src # REQUIRED for Tier 1+ exploitation_tier: 2 # 1=pattern-only, 2=payload-test, 3=full-exploit vuln_types: # Which pipelines to run - injection # SQL, NoSQL, command injection - xss # Reflected, stored, DOM XSS - auth # Auth bypass, session, JWT - ssrf # URL scheme abuse, metadata max_cost_usd: 15 # Budget cap per run timeout_minutes: 30 # Time cap per run require_authorization: true # MUST confirm target ownership no_production: true # Block production URLs production_patterns: # URL patterns to block - "*.prod.*" - "api.*" - "www.*"
Safeguards (Mandatory)
Authorization Gate
Every pentest validation run MUST:
Display target URL and exploitation tier to user
Require explicit confirmation: "I own/authorized testing of this target"
Log authorization with timestamp
Block if target URL matches production patterns
What This Skill Does NOT Do
Full autonomous reconnaissance (Nmap, Subfinder)
Zero-day exploit development
Attack targets without explicit authorization
Test production systems
Store actual exfiltrated data (only proof of access)
"No Exploit, No Report." A vulnerability scanner that can't prove exploitation delivers uncertain value. This skill transforms security findings from theoretical risks into proven vulnerabilities with evidence. Every confirmed finding comes with a reproducible proof-of-concept. Every false positive is eliminated before it reaches the report.
Think proof, not prediction. Don't report what MIGHT be vulnerable. Prove what IS vulnerable.
Installed into a catalogue, chosen by a router
Install qe-pentest-validation and it becomes one more candidate the router can pick - when it fits.
A listing is a routing target with a manifest: what it does, which domains it covers, what it costs and who publishes it. Once installed it sits beside your own models and tools, is scored like any other candidate for each request, and shows up in the trace when it wins. Ratings come from workspaces that installed it, one per account.
When a provider is down or over quota the request moves on; the trace says it did.
Use it
Copy one of these into your project. Installing also returns the manifest and these snippets.
# after Install: the listing is in your workspace's routing pool - a plan picks it for its slotcurl-s-X POST https://api.opensmartroute.ai/api/v1/route -H 'Authorization: Bearer $OSR_API_KEY' -H 'Content-Type: application/json' -d '{"text": "...", "plan": true}'
Manifest
An Open Capability Manifest: the router reads it to know what this does, what it costs and when to pick it.