AI Authorship Disclosure
What this page covers
Stigmem is built by two contributors with heavy AI-coding assistance. We disclose this because a category whose product is trust should not quietly hide where the work came from.
This is a calibration aid, not a defect notice.
It helps readers decide where to apply extra verification before relying on Stigmem in their own workload.
Deeper human review
These paths receive line-by-line human review:
spec/
Protocol specification text.
docs/adr/
Architecture Decision Records.
Top-level repo docs
LIMITATIONS.md, SECURITY.md, MAINTAINERS.md, and the root README.md.
Threat-model entries
In spec/security/ and rendered security docs.
Lighter human review
These paths receive high-level direction and spot checks:
node/src/
Reference node implementation.
adapters/
Adapter implementations.
sdks/
SDK stubs.
UI scaffolding
And experimental surfaces.
Test suites
Non-spec docs
Outside spec, ADRs, and security posture.
Contributor expectation
When a PR includes AI-generated code or prose, say so in the PR description.
The project does not reject AI-assisted contributions; it calibrates review effort. Reviewers should verify behavior against the spec, run the relevant conformance or test suite, and audit higher-risk paths before merging.
The canonical source copy for this disclosure is also maintained in README.md and CONTRIBUTING.md.