Commit graph

320 commits

Author SHA1 Message Date
twentyOne2x
a3cce93d33 Guide Leo through evidence-backed claim challenges 2026-07-15 21:23:30 +02:00
twentyOne2x
56ad1c41dc
Merge pull request #165 from living-ip/codex/leo-observatory-private-staging-t3-w2-20260715
Some checks are pending
CI / lint-and-test (push) Waiting to run
Restore Observatory staging federation route
2026-07-15 21:22:51 +02:00
twentyOne2x
a30f48415a Bind Leo responses without masking semantic failures 2026-07-15 21:21:17 +02:00
twentyOne2x
8e198a791b Ground Leo reasoning in bounded live retrieval 2026-07-15 21:21:06 +02:00
twentyOne2x
b7197d6719 Restore Observatory staging federation route 2026-07-15 21:19:59 +02:00
twentyOne2x
9cd4a830cc Restore read-only GCP IAP status route 2026-07-15 21:19:59 +02:00
twentyOne2x
b6efa6c4a3 Keep challenger DB limits compatible with OOS retrieval 2026-07-15 13:10:15 +02:00
twentyOne2x
4d3e4afe21 Merge remote-tracking branch 'origin/pr-160' into HEAD 2026-07-15 13:04:29 +02:00
twentyOne2x
9a5305f386
Merge pull request #152 from living-ip/codex/leo-agent-challenger-loop-20260715
Some checks are pending
CI / lint-and-test (push) Waiting to run
Prove isolated Leo agent-challenger revision loop
2026-07-15 12:56:39 +02:00
Codex Integration
822438bd2f Merge main and repair GCP runtime integration 2026-07-15 12:51:17 +02:00
twentyOne2x
8a4325f42a Harden OOS benchmark provenance and leakage checks 2026-07-15 12:06:35 +02:00
twentyOne2x
1c6b7b182e Pin lifecycle receipts to the reviewed parity manifest 2026-07-15 10:59:43 +02:00
twentyOne2x
52fc0a88fe Normalize database authorization parity and validate hashes 2026-07-15 10:49:55 +02:00
twentyOne2x
cbcb293c10 Merge remote-tracking branch 'origin/main' into codex/leo-agent-challenger-loop-20260715 2026-07-15 10:34:04 +02:00
twentyOne2x
f461f98b47 fix natural causal objections and retain failure traces 2026-07-15 10:33:27 +02:00
twentyOne2x
1a3305ac5d Merge remote-tracking branch 'origin/main' into codex/leo-gcp-data-parity-20260715
# Conflicts:
#	.agents/skills/teleo-gcp-parity-ops/SKILL.md
2026-07-15 10:26:26 +02:00
twentyOne2x
1a043d36a1 Harden GCP parity proof against forged lifecycle claims 2026-07-15 10:20:27 +02:00
twentyOne2x
36fe0d3cc2
Harden Observatory private staging deploy gate (#163)
Some checks are pending
CI / lint-and-test (push) Waiting to run
Merge the T2 readiness and least-privilege staging gate. This does not deploy Cloud Run, mutate Cloud SQL, or change production routing.
2026-07-15 09:55:49 +02:00
twentyOne2x
959ba3cb7b test: wait for Linux service environment readiness 2026-07-15 09:51:51 +02:00
twentyOne2x
e172207465 preserve fetched challenger receipts through projection failures 2026-07-15 09:11:53 +02:00
twentyOne2x
1b21139d4e test: mirror runtime trust environment 2026-07-15 07:35:59 +02:00
twentyOne2x
25c2682476 test: isolate runtime guard from CI environment 2026-07-15 07:31:37 +02:00
twentyOne2x
0de581c921 Merge remote-tracking branch 'origin/main' into codex/leo-agent-challenger-loop-20260715 2026-07-15 07:27:07 +02:00
twentyOne2x
d5cc440acf fix challenger semantic and whole-report gates 2026-07-15 07:26:58 +02:00
twentyOne2x
8490a0d8c8 Merge remote-tracking branch 'origin/main' into codex/leo-gcp-runtime-security-20260715 2026-07-15 07:16:48 +02:00
twentyOne2x
7b16c6d3cc fix: prove least-privilege GCP Leo runtime context 2026-07-15 07:16:40 +02:00
twentyOne2x
33399bd9ac
Merge pull request #153 from living-ip/codex/leo-telegram-visible-chain-20260715
Some checks are pending
CI / lint-and-test (push) Waiting to run
Prepare zero-mutation Telegram-visible Leo reasoning chain
2026-07-15 06:28:23 +02:00
twentyOne2x
93eddbcd4f Model conditional no-DB receipt gaps 2026-07-15 06:23:15 +02:00
twentyOne2x
35c9531950 fix: make local Telegram verification preparation-only 2026-07-15 06:17:20 +02:00
twentyOne2x
d089fd9804 Merge remote-tracking branch 'origin/main' into codex/leo-proposal-apply-lifecycle-20260715 2026-07-15 05:58:40 +02:00
twentyOne2x
8506dd9359 Bind Leo responses without masking semantic failures 2026-07-15 05:54:45 +02:00
twentyOne2x
224a448bb1 Verify retained live-schema lifecycle proof 2026-07-15 05:34:16 +02:00
twentyOne2x
da65e05316 Ground Leo reasoning in bounded live retrieval 2026-07-15 05:30:10 +02:00
twentyOne2x
0125fbff12 Use composite evidence keys in rollback dependencies 2026-07-15 05:28:59 +02:00
twentyOne2x
8ee04f1cda Test composite evidence rollback identity 2026-07-15 05:25:19 +02:00
twentyOne2x
000bda06c2 Apply evidence against the live composite schema 2026-07-15 05:20:34 +02:00
twentyOne2x
ed827626db Support legacy claim evidence fingerprints 2026-07-15 05:14:47 +02:00
twentyOne2x
604e1669c7 Harden proposal apply and rollback provenance 2026-07-15 05:10:04 +02:00
twentyOne2x
b238fd6fc2 fix: redact private paths from identity receipts 2026-07-15 04:45:57 +02:00
twentyOne2x
8f87c540dc fix: harden Telegram proof chain against forged evidence 2026-07-15 04:41:00 +02:00
twentyOne2x
d299f1cdf5 Merge remote-tracking branch 'origin/main' into codex/leo-gcp-runtime-security-20260715 2026-07-15 04:40:36 +02:00
twentyOne2x
b7bdcea46a fix: enforce scoped GCP Leo runtime credentials 2026-07-15 04:40:26 +02:00
twentyOne2x
3afd1dbb5e feat: reproduce Leo identity across disposable restarts 2026-07-15 04:35:15 +02:00
twentyOne2x
626bac493e
Merge pull request #151 from living-ip/codex/leo-deterministic-db-rebuild-20260715
Some checks are pending
CI / lint-and-test (push) Waiting to run
Rebuild exact revise-strategy database transitions
2026-07-15 04:17:35 +02:00
twentyOne2x
14a7529a90
Merge pull request #149 from living-ip/codex/leo-source-ingestion-proposals-20260715
Some checks are pending
CI / lint-and-test (push) Waiting to run
Stage source-linked ingestion proposals without canonical writes
2026-07-15 04:11:33 +02:00
twentyOne2x
37c0ed091d feat(ops): bind snapshots to private GCP parity lifecycle 2026-07-15 04:11:29 +02:00
twentyOne2x
b99f82bb6b fix: make challenger proof retention-safe 2026-07-15 04:09:07 +02:00
twentyOne2x
2ad14a915d Fix live benchmark preflight embedding 2026-07-15 04:04:33 +02:00
twentyOne2x
ae117097af Add blinded live Leo reasoning benchmark 2026-07-15 04:02:41 +02:00
twentyOne2x
bd3c17628c docs: retain production apply authorization boundary 2026-07-15 04:01:29 +02:00
twentyOne2x
40344c91d7 enforce revise strategy approval timestamp provenance 2026-07-15 04:00:53 +02:00
twentyOne2x
3b138fd2e1 fix: bind ingestion acceptance to exact provenance 2026-07-15 04:00:14 +02:00
twentyOne2x
db64500a12 feat: rehearse approved proposal apply and rollback 2026-07-15 04:00:09 +02:00
twentyOne2x
92da54d804
Add authenticated GCP Observatory read adapter (#154)
Some checks are pending
CI / lint-and-test (push) Waiting to run
2026-07-15 03:43:05 +02:00
twentyOne2x
dfa6be6ba0 feat: prepare zero-mutation Telegram Leo chain proof 2026-07-15 03:37:53 +02:00
twentyOne2x
45cf14a1df Merge remote-tracking branch 'origin/main' into codex/leo-agent-challenger-loop-20260715 2026-07-15 03:35:06 +02:00
twentyOne2x
af5af06130 test: preserve concurrent-runtime rejection and harden semantics 2026-07-15 03:23:34 +02:00
twentyOne2x
a6947a3f8b replay exact revise strategy database transitions 2026-07-15 03:20:02 +02:00
twentyOne2x
15c30f1fbe
Make Leo skill onboarding clean-context reproducible (#150)
Some checks are pending
CI / lint-and-test (push) Waiting to run
* make Leo skill onboarding clean-context reproducible

* harden Leo clean-context skill acceptance
2026-07-15 03:19:15 +02:00
twentyOne2x
6a7f2d3f8c fix: keep challenger revision free of compiled context 2026-07-15 03:18:28 +02:00
twentyOne2x
56b0b899b1 test: preserve rejected challenger receipt and harden replay 2026-07-15 03:11:54 +02:00
twentyOne2x
586c215b54 fix: harden Leo challenger proof gates 2026-07-15 02:54:05 +02:00
twentyOne2x
1e58b44fae feat: stage source-linked ingestion proposals without canonical writes 2026-07-15 02:46:18 +02:00
twentyOne2x
2da35f929e feat: add isolated Leo challenger runtime 2026-07-15 02:40:53 +02:00
twentyOne2x
68625b8afa Merge remote-tracking branch 'origin/main' into codex/leo-gcp-runtime-security-20260715 2026-07-15 02:18:01 +02:00
twentyOne2x
af4f18fddc Harden unseen Leo benchmark against valid reasoning variance 2026-07-15 01:15:06 +02:00
twentyOne2x
678154da9f Prove unseen Leo claim challenge and revision chain 2026-07-15 00:50:01 +02:00
twentyOne2x
0cc05aa2f4 Preserve natural claim follow-up reasoning 2026-07-15 00:42:33 +02:00
twentyOne2x
1a6b7d51a4 Add deterministic genesis plus ledger rebuild 2026-07-15 00:33:29 +02:00
twentyOne2x
4088c9c8f5 Harden Leo turn attribution and no-write gate 2026-07-15 00:33:16 +02:00
twentyOne2x
c2eebb9654 fix: fail closed on missing DB fingerprints 2026-07-15 00:06:33 +02:00
twentyOne2x
cc46d66713 feat: bind tested Leo turns to reproducible inputs 2026-07-14 23:56:33 +02:00
ce90a79aac fix: scope GCP Leo database runtime 2026-07-14 12:57:31 -07:00
twentyOne2x
a5a4163a2b docs: record GCP deploy and restart proof 2026-07-14 12:54:30 +02:00
twentyOne2x
8451c51087 fix: keep GCP knowledge reads on Cloud SQL 2026-07-14 12:49:12 +02:00
twentyOne2x
9207e1b100 feat: prove database-first Leo on private GCP copy 2026-07-14 12:32:26 +02:00
twentyOne2x
56a99743fc Record live Leo database reasoning proof 2026-07-14 10:12:16 +02:00
twentyOne2x
9d4baafc69 Prove Leo database tool attribution 2026-07-14 10:01:47 +02:00
twentyOne2x
ad70bd4876 Enforce Leo reply budgets before delivery 2026-07-14 09:46:56 +02:00
twentyOne2x
281b57ef53 Harden Leo DB retrieval and behavior attribution 2026-07-14 09:31:49 +02:00
Fawaz
6483c7b164
Add protected canonical claims read endpoint for the Observatory (#139)
Some checks are pending
CI / lint-and-test (push) Waiting to run
* Add protected canonical claims read API

* Normalize claim loader exits in worker thread

* Allow file-gated claims reader defaults
2026-07-14 00:18:14 -04:00
twentyOne2x
7ea07cc4fc retain replay receipts for guarded KB applies 2026-07-14 03:26:42 +02:00
twentyOne2x
16aa3f41b9 fix named proposal response validation 2026-07-14 02:23:06 +02:00
twentyOne2x
5e7625d985 fix Leo named proposal readback routing 2026-07-13 22:40:06 +02:00
twentyOne2x
ee3ee1d998 Preserve atomic claim text in source proposals 2026-07-13 22:13:19 +02:00
twentyOne2x
dd724992f6 Use dense source references for document extraction 2026-07-13 22:05:31 +02:00
twentyOne2x
ab39407072 Make source preparation preserve exact document text 2026-07-13 21:58:05 +02:00
twentyOne2x
e304e1f296 prepare DB-grounded source proposals from documents 2026-07-13 21:47:51 +02:00
twentyOne2x
d9063c4c57
prevent personal-name leakage in Leo bootstrap (#126)
Some checks are pending
CI / lint-and-test (push) Waiting to run
2026-07-13 21:25:47 +02:00
twentyOne2x
5dfcd32c34
fix Leo broad source intake truth (#124)
Some checks are pending
CI / lint-and-test (push) Waiting to run
2026-07-13 20:59:28 +02:00
twentyOne2x
ee48f82f06
Stage source artifacts as reviewable KB proposals (#123)
Some checks are pending
CI / lint-and-test (push) Waiting to run
2026-07-13 20:38:18 +02:00
twentyOne2x
9a76f480c5
Include approved rows in Leo demo readback (#122)
Some checks are pending
CI / lint-and-test (push) Waiting to run
2026-07-13 20:17:29 +02:00
twentyOne2x
73f466f899
Make broad Leo replies deterministic (#121)
Some checks are pending
CI / lint-and-test (push) Waiting to run
2026-07-13 20:07:51 +02:00
twentyOne2x
5d135d8818
Make broad Leo answers database-composed (#120)
Some checks are pending
CI / lint-and-test (push) Waiting to run
2026-07-13 19:53:25 +02:00
twentyOne2x
ec64dbf6d0
Fail closed on unavailable Leo database truth (#119)
Some checks are pending
CI / lint-and-test (push) Waiting to run
* Fail closed on unavailable Leo database truth

* Cover direct DB intents in fail-closed fallback

* Harden broad Leo database intent routing

* Fail closed on incomplete matrix state

* Distinguish unobserved matrix state
2026-07-13 14:36:30 +02:00
twentyOne2x
b6408cdbd1
Compose broad Leo answers from live database state (#118)
Some checks are pending
CI / lint-and-test (push) Waiting to run
Validated by 1121 repository tests and a read-only six-question smoke against live VPS Postgres before installation.
2026-07-13 13:59:16 +02:00
twentyOne2x
61c1fa9165
Accept explicit source-intake prohibitions (#117)
Some checks are pending
CI / lint-and-test (push) Waiting to run
2026-07-13 13:35:34 +02:00
twentyOne2x
2264ad1f2b
Enforce DB-composed Leo responses before delivery (#116)
Some checks are pending
CI / lint-and-test (push) Waiting to run
2026-07-13 13:27:54 +02:00
twentyOne2x
144e619eee Keep Leo composition replies within budget 2026-07-13 12:58:24 +02:00
twentyOne2x
033e251fd0 Disambiguate governance rules in Leo composition 2026-07-13 12:52:50 +02:00