Palimpsest
by beepboop2025communityv1.0.0
3
Palimpsest is an append-only, verifiable ledger for two public records: AI evaluation results and observed content deletions from public online sources. It lets anyone verify offline that published entries were not altered, reordered, or dropped after publication.
From the registry: Live internet-censorship signals: Great Firewall reachability, model-layer censorship, takedowns.
Install & configure
No canonical install info was extracted for this server. Check the repository README directly.
Tools (5)
verify_eval_registry.py
Verifies the AI eval registry hash chain and checks that probe sets were pre-registered before model runs.
verify_ledger.py
Verifies the censorship/erasure ledger for integrity.
prove_inclusion.py
Generates or checks an inclusion proof for a single sealed result against the Merkle root.
validate_ddti.py
Validates the censorship scoring method against documented events.
palimpsest_demo.py
Runs a demo that pulls live censorship feed data and ranks what the censor is focused on; can also run offline with sample data.
Prerequisites
- • python3
Remote Endpoints
https://api.seiche.info/palimpsest/mcp
type: streamable-http
Details
- Language
- Python
- License
- MIT
- Released
- Jul 14, 2026
- Last commit
- Jul 19, 2026
- Registry name
- io.github.beepboop2025/palimpsest