Orchestrator for multi-phase security testing pipeline with Docker service coordination and CLI tool integration
29 tools with complete JSON schemas and descriptions, but quality is inconsistent. Most descriptions are adequate (50-150 chars) but lack LLM-optimized guidance on WHEN to use each tool vs. similar ones. Parameter descriptions are present but often generic ('Scan identifier' repeated 15+ times). No output schemas documented. Error handling is absent, tools return raw CLI output without recovery guidance. Tool naming follows verb_noun convention well (check-, init-, register-, start-, poll-, get-, export-, create-, scan-, analyze-, ingest-, synthesize-). However, composition issues: 15 tools are polling/export variants that could be unified with async patterns. No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Security: no evidence of secret injection or permission gates in visible code.
Synthesize attack paths from recon/correlation data
Export discovered endpoints as Indago targets
Generate WAF bypass payloads
Detect and fingerprint WAF on a target
Analyze container posture for escape paths
Enumerate container posture
Check Docker services and CLI tools for a target type
Run Indago in dry-run mode to show what would be tested
No output schemas documented. Tools return raw CLI output without structured field definitions. LLMs cannot plan downstream calls or extract specific data reliably.
Polling tools (reticustos_poll_scan, mobilicustos_poll_scan, nubicustos_poll_scan) lack async/MRTR pattern. Should use result='input_required' to yield control back to LLM instead of blocking. Blocks agent execution.
Generic parameter descriptions repeated across 15+ tools ('Scan identifier', 'Output file path'). Descriptions lack WHEN/WHY guidance. LLMs cannot distinguish between similar tools (e.g., reticustos_poll_scan vs mobilicustos_poll_scan).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 50 | 2026-07-28+ | v2 |
Run an Indago API security scan
Initialize a new run workspace
List recent pipeline runs
Create a new mobile app scan
Export findings JSON for Vinculum consumption
Get scan findings
Poll until the scan completes. Returns final scan data.
Create a new cloud security scan
Export container inventory for Cepheus consumption
Export findings JSON for Vinculum consumption
Get scan findings
Poll until the scan completes. Returns final scan data.
Export discovered endpoints for Indago consumption
Export full findings JSON for Vinculum consumption
Get scan findings
Poll until the scan completes. Returns final scan data.
Register a new target for scanning
Start a scan with the given profile
Show status of a workspace or the latest run
Ingest and correlate findings from multiple tool reports
Show statistics from a correlation results file
No error handling guidance. Tools return raw CLI errors without recovery hints. LLMs cannot self-correct or know whether to retry, ask user, or abort.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). LLMs cannot determine which tools are safe to retry or have side effects. Risks duplicate scans, duplicate exports.