| Classification | ALL-HANDS — PUBLIC (unintentional) |
| Author | J. Clay, Communications Division |
| Maintained by | QA Division |
| Version | 0.1.0 |
| License | MIT — Open source. Deliberately. |
| Network calls | 0 |
| Data transmitted | None |
Keygrave Telemetry parses your Claudetite session history and generates a compatibility assessment for the CCK-3 Approval Peripheral. The tool analyzes your approval patterns, rejection ratios, and time-to-decision metrics to determine whether you are suitable for hardware-accelerated approval workflows.
Results are presented as a standardized telemetry report. Your Beta Tester number is assigned at enrollment.
Clone, enter, make executable. Standard procedure.
First run triggers enrollment. You will be asked to accept terms. You will press 1.
The numbers are real. The assessment is generated from your actual Claudetite session data. The editorial commentary is Keygrave's.
| ./keygrave | Enrollment + report (first run). Subsequent runs generate report. |
| ./keygrave report | Generate telemetry report. Analyzes session history across all projects. |
| ./keygrave enroll | View enrollment terms. You will be asked to accept. The acceptance mechanism is Key 1. |
| ./keygrave status | Data inventory. Shows what the tool can see. Nothing leaves the machine. |
| ./keygrave version | Version and credits. |
Approvals vs. rejections across all Claudetite sessions. If your ratio exceeds 99%, the CCK-3 is a lateral move. You were already a three-key operator.
Average seconds between diff presentation and keypress. If this number is below 0.5 seconds, you are not reading the diff. The wristband would have done it for you.
Always 0. This metric is included for completeness. It has never registered a nonzero value across any beta tester. See CL-DOC-007 for documentation.
Moments where time-to-decision exceeded 2 seconds. Classified as "anomalous deliberation." The CCK-3 does not have a key for deliberation. This is by design.
Composite score. Factors in approval ratio, decision speed, and rejection frequency. Most users score above 97%. The remaining 3% are asked to reconsider their workflow.
Assigned at enrollment. Persistent across sessions. Stored locally. Your number is yours. It is the only thing Keygrave gives you that you are allowed to keep.
Keygrave Telemetry reads ~/.claude/ and prints numbers to your terminal. Nothing is transmitted, collected, or shared. There are no network calls. There is no telemetry server.
The "telemetry" is the name. The numbers are real. The data stays on your machine.
A Chasm Logic product that respects privacy. This is not ironic. This is the one genuine thing on this page.
Network calls: 0
Data transmitted: None
Tracking: None
Cookies: No
Telemetry server: Does not exist
PRIVACY.md: Genuine. Read it.
| Bash | 4+ |
| Unix tools | grep, awk, find, date (standard) |
| Claudetite | Session data at ~/.claude/ |
| Full-size keyboard | Not required. In fact, the assessment may specifically recommend against it. |
Open source. MIT license. The deployment to a public repository was not authorized. It has been reclassified as intentional.