Capability inventory
September 12, 2026 · View on GitHub
These counts describe this source revision, not necessarily the latest PyPI release. CI derives each count from the code and checks this table. Counts are an inventory, not a benchmark or a measure of detection quality.
| Surface | Count |
|---|---|
| Probes in the library | 65 |
| HTTP probes | 64 |
| Oracle kinds | 28 |
| Tool wrappers | 203 |
| …that parse output into findings today | 20 |
| MCP tools | 52 |
| Specialist agents | 18 |
Registered probes do not all have verification recipes. A wrapper without an output parser may return raw tool output without creating structured findings. Neither an agent count nor an MCP tool count implies coverage of every target type. See the verification design and versioned benchmark reports for evidence.