Skip to main content
For SDK maintainers and indexer operators who run the corpus or the tools in their own CI. It lists what could go wrong with the kit’s data and code, which control covers each threat and where that control lives, what risk is left, and what is watched after a release.

Scope

The kit is an offline test kit: command-line tools, data files and CI workflows. It runs no service, holds no user data and touches no funds. What needs protecting:
  1. The integrity of the corpus and its expected values. A wrong byte or a wrong expectation makes a correct reader fail or lets a buggy one pass.
  2. The integrity of what downstream CI executes. Other projects run the corpus, and some run the tools, in their own CI.
  3. The confidentiality of unreported security defects the kit finds in other projects, until those projects have been told.

Threats and mitigations (STRIDE)

Each mitigation names where it is implemented or checked. Residual risk is stated plainly.

Monitoring plan

There is no production system to monitor: no uptime, no user metrics. What is monitored is the corpus’s integrity, upstream drift and the state of findings. “In place” means the job or check exists and runs today. A nightly failure turns the scheduled run red, writes the failed job and the command to rerun it to the run’s job summary, and GitHub notifies the workflow’s owner. No secret or external service is involved. Opening an issue from a failure is a maintainer’s step.