Project-maintained results linked from the benchmark map and dated verification ledgers.
Community reports are signals, not automatic claims.
Share what you tested, how you tested it, what changed, and what the result does not prove. Good field reports make future evidence stronger without asking anyone to expose private source.
Category status
A dedicated GitHub Discussions category named Evidence & Field Reports has been requested. Until it exists, use Show and tell with a [Field Report] title.
What goes where.
Real runs, screenshots, logs, demos, surprises, and reproduction notes from users.
Useful product-feel evidence when the fixture, setup, and narrow claim are visible.
Failures, non-reproducible runs, unsupported environments, and claims that still need controlled proof.
Use this shape.
## What I tested
## Environment
- OS:
- AIppocampus version / commit:
- Host agent:
- Model:
- Dataset / thread type:
## Steps
## Result
## What surprised me
## Reproducibility
- Can others reproduce this? yes/no/partial
- Any private data removed? yes/no
## Claim boundary
This shows:
This does not show:Do not post private raw conversations, local paths, credentials, cookies, tokens, or private registry exports. Redacted excerpts and public-safe fixtures are enough.
Official evidence still has one map.
The public evidence entrypoint should make community results easy to find without creating a second source of truth. Official claims still flow through the benchmark evidence map, readiness snapshot, and dated verification ledger.