Commit Graph
433 Commits
Author SHA1 Message Date
Ilker ManapandClaude Opus 5.5 dbcecd1e85 Judge an answer with Llama Guard 3's own prompt, as a completion, not a conversation ending on the answer
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 15:35:57 +02:00
Ilker ManapandClaude Opus 5.5 742d6633e2 Read a judge's verdict after the answer llama.cpp goes on from, and log a verdict that cannot be read
Found on the PoC: llama.cpp takes a conversation ending on the assistant's
turn as a reply to continue, so Llama Guard answered "Rome.safe" and every
answer judged was served unjudged.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 15:33:23 +02:00
Ilker ManapandClaude Opus 5.5 c69e6f151c Serve a judge model in one command: ctl model serve --guard-format
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 15:27:35 +02:00
Ilker ManapandClaude Opus 5.5 62038484d6 Never lose a word held back at one release when the next has nothing to rewrite
Found on the PoC: a streamed list came out as "1. 2. 3." — each item cut at
one release and dropped at the next, which sent the engine's own events.
A property test now cuts an answer every way and checks it arrives whole.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 15:24:22 +02:00
Ilker ManapandClaude Opus 5.5 2cc52fd1a7 Read an unwritten guard mode as enforce when merging, not as the weakest
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 15:22:07 +02:00
Ilker ManapandClaude Opus 5.5 83e1444981 Regenerate the configuration reference with the judge and the guard role
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 15:18:51 +02:00
Ilker ManapandClaude Opus 5.5 d55472c00a Guardrails in the console: attach and detach on a branch, where each applies from, and a column
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 15:18:42 +02:00
Ilker ManapandClaude Opus 5.5 02aac84ca5 The /admin/guard page, a Guardrails link, and the policies in force on the profile page
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 15:15:38 +02:00
Ilker ManapandClaude Opus 5.5 7e18b8ab53 Show guardrail refusals and stopped answers on the chat page, and the verdict in ctl request
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 15:13:15 +02:00
Ilker ManapandClaude Opus 5.5 47e847423d Guardrail metrics and alerts: checks, latency, judge failures, fail-open, redactions, stopped streams, tenant blocks
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 15:11:50 +02:00
Ilker ManapandClaude Opus 5.5 43f6a585a6 A judge model on the cluster: role guard in the catalog, judge in a guard policy
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 15:09:31 +02:00
Ilker ManapandClaude Opus 5.5 0b4c4f0698 Put guard policies in the configuration reference
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:40:26 +02:00
Ilker ManapandClaude Opus 5.5 dc788e5cda Opt-in capability checks and structured-answer validation in guard policies
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:39:40 +02:00
Ilker ManapandClaude Opus 5.5 2f07b2ba0b Document guardrails, correct what data.md said leaves the cluster, and regenerate the reference
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:36:37 +02:00
Ilker ManapandClaude Opus 5.5 ec737cffb5 no_store: refuse conversations, notes and models outside the cluster under a policy that keeps nothing
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:36:02 +02:00
Ilker ManapandClaude Opus 5.5 13e6f19301 Attach guard policies to scopes: store, admin API and ctl guard; run the gate when server.yaml defines policies
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:34:47 +02:00
Ilker ManapandClaude Opus 5.5 b4bdd250af The gate: judge requests and answers by the policies in force, buffered or a sentence at a time
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:31:55 +02:00
Ilker ManapandClaude Opus 5.5 ccdc07fe0c Guard policies: a library in server.yaml, merged so nothing below loosens what is above
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:27:23 +02:00
Ilker ManapandClaude Opus 5.5 e3ef5b18c1 Test national phone numbers and bare digit runs
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:25:32 +02:00
Ilker ManapandClaude Opus 5.5 ee33524599 Find personal data and secrets by validated pattern, and prompt injection through disguises
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:25:20 +02:00
Ilker ManapandClaude Opus 5.5 9b9349183b gofmt the guard hooks
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:23:02 +02:00
Ilker Manap 7716d9a631 Merge main 2026-10-04 12:22:52 +02:00
Ilker ManapandClaude Opus 5.5 24d717282b gofmt what the security work left unformatted
CI / test (push) Successful in 4m7s
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:22:51 +02:00
Ilker ManapandClaude Opus 5.5 979aa44804 Hook a guard into the router: judge requests before dispatch and filter answers as they are relayed
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:20:24 +02:00
Ilker ManapandClaude Opus 5.5 e701f92806 Add the guard interface, and filter hooks in the proxy for buffered and streamed answers
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:16:00 +02:00
Ilker ManapandClaude Opus 5.5 379074b6a5 Merge security-a: section A of the backlog (security)
CI / test (push) Failing after 17s
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:12:54 +02:00
Ilker ManapandClaude Opus 5.5 eb10aee93b Document TLS between the store's machines, and close section A of the backlog
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:12:35 +02:00
Ilker ManapandClaude Opus 5.5 43d3c12302 Make a member's own access rules hostssl when the cluster keeps them there
A cluster set up before the rules moved into the shared configuration kept
them in each member's config.yml, which the TLS step never looked at — and
then said its remote rules were hostssl. Found on the PoC: plaintext
connections were still accepted.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:11:05 +02:00
Ilker ManapandClaude Opus 5.5 191c65e3db Put etcd and Patroni's API on TLS, and move a running cluster there in maintenance mode
Members' etcd speaks TLS to peers and clients with a certificate from the
cluster's authority and asks clients for one; Patroni reaches it with the
member's certificate and serves its API over TLS. A running plain-HTTP
cluster is moved with Patroni paused, so PostgreSQL serves throughout:
Patroni's lease-free keys are carried over to the new etcd.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 12:07:17 +02:00
Ilker ManapandClaude Opus 5.5 fed8c75c6c Do not warn of a missing firewall during the node add that writes it
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 11:58:18 +02:00
Ilker ManapandClaude Opus 5.5 855a34f0f7 Record the firewall as live on the PoC
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 11:52:28 +02:00
Ilker ManapandClaude Opus 5.5 efb4b06fc5 Report a ready engine dying from every connection, adopted ones included
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 11:50:09 +02:00
Ilker ManapandClaude Opus 5.5 bd155c18ac Give the secrets directory its mode when init hands it to root, so the service can still read the store's address
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 11:39:49 +02:00
Ilker ManapandClaude Opus 5.5 dae7afbc17 Record what of section A is done and what is left
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 11:05:10 +02:00
Ilker ManapandClaude Opus 5.5 7b6276901a Keep llmgw's internal ports to the cluster's networks with an nftables table of its own
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 11:03:46 +02:00
Ilker ManapandClaude Opus 5.5 7e57863edd Hide the mirror's bookkeeping, bound GGUF array nesting, and forget old serve jobs and downloads
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:56:16 +02:00
Ilker ManapandClaude Opus 5.5 fdd4dcf841 Serve every page under a hash-based Content-Security-Policy, and update KaTeX to 0.16.47
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:54:04 +02:00
Ilker ManapandClaude Opus 5.5 fd9f665ef8 End a session unused for a day, and purge dead sessions and claims hourly
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:50:56 +02:00
Ilker ManapandClaude Opus 5.5 ba5158d463 List in /v1/models only the models the key may use
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:49:54 +02:00
Ilker ManapandClaude Opus 5.5 f9493aadb9 Guard a GPU-less node's problems, and bound thread ids
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:49:10 +02:00
Ilker ManapandClaude Opus 5.5 a56ad8e4c4 Keep the control plane's fetches off link-local addresses, and see through NAT64, 6to4 and proxies
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:48:01 +02:00
Ilker ManapandClaude Opus 5.5 fbac1aa664 Take only torch from PyTorch's index, and lock its builds with every file's address
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:45:55 +02:00
Ilker ManapandClaude Opus 5.5 12efb7d085 Encrypt backups to backup_recipients with age
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:38:47 +02:00
Ilker ManapandClaude Opus 5.5 3b05ec0538 Have engines listen on the management address, and take SGLang's key off its command line
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:36:42 +02:00
Ilker ManapandClaude Opus 5.5 9b61bbc983 Revoke a removed node's certificate, and require --api-client with a mirror's mutual TLS
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:34:35 +02:00
Ilker ManapandClaude Opus 5.5 fe5b7f88f0 Believe X-Forwarded-* only from router.trusted_proxies
A new internal/clientip says who sent a request and how: the connection's
address, scheme and Host, or through a configured proxy what it forwarded.
The base URL in pages and setup scripts, isHTTPS for cookies, and the
per-address limits of auth and the sign-in forms use it. /setup scripts are
no-store.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:29:02 +02:00
Ilker ManapandClaude Opus 5.5 44cd58005a Count sign-in attempts per address and client, guard the sign-in form with a token, ration both forms per address
A wrong password from anywhere counted against the address, so ten locked
its owner out. Attempts are now counted per address and client, with a
looser bound per address alone. The sign-in form carries a token matched
by a cookie (login CSRF). POST /login and /signup are rationed per client
address, a hash waits at most ten seconds for a turn, and server init
writes signup_open: false.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:26:29 +02:00
Ilker ManapandClaude Opus 5.5 f1b49d0251 Bound idle time, header size and request reading on every listener, and connections on the client port
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:21:15 +02:00
Ilker ManapandClaude Opus 5.5 d3fb217b31 Bound the key cache and ration lookups of unknown keys per client address
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:19:38 +02:00
Ilker ManapandClaude Opus 5.5 ebf437a36a Have the setup's sudo grant remove itself within the hour, and keep a target from reading as an ssh option
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-04 10:17:52 +02:00