Improve README and add opt-in Jev workflows - #252
Coding-Dev-Tools wants to merge 59 commits into
Conversation
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 71c099dfd8
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 9165f7847f
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 3bec5bc6ed
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 24f0fd4973
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: cf6b4268ae
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: d299855513
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 6ab0475f13
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 0f6d292beb
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 815edc9765
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 94b8d244bf
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: fd5b060adb
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| "engraphis/core/poisoning.py": "5bc67169ee8032f3777f2d3969dcf71821437bd473ce4f917c4b41a50e845fb6", | ||
| "engraphis/core/query_planner.py": "249062d67392ab7c203cc71e9040e99bee91bf570604e90949149a93cb652120", | ||
| "engraphis/core/read_snapshots.py": "be08e63a88bd38ed91d61b28657a65201c38994856db2798b209a73151dd202a", | ||
| "engraphis/core/recall.py": "92e3e8a01101c21b8b1179abb8d5357e33f808f190b84d83a405cda92d28b04d", |
There was a problem hiding this comment.
The current v148 registry records stale hashes for engraphis/core/recall.py and engraphis/service.py: running python -m pytest tests/ -q -x fails test_public_numeric_evidence_registry_is_complete_and_live because the recorded hashes (92e3e8a… and e4a4d3e…) differ from the committed files (ce7ac0e… and 6e36dbe…). This is fresh evidence beyond the earlier v131 comment: the final fixes changed a different pair of source files after v148 was generated, so regenerate the artifact and update its checksum consumers to restore the required CI gate.
AGENTS.md reference: AGENTS.md:L50-L54
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 062bff700d
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if selected_decision_backend in {"managed", "auto"}: | ||
| # Keep the managed client lazy: a dashboard login may arrive after | ||
| # engine construction, and configuration checks happen per call. | ||
| jev_client = EngraphisCloudDecisionClient() |
There was a problem hiding this comment.
When ENGRAPHIS_DECISION_BACKEND is managed or auto and a signed-in caller enables jev_assisted with allow_remote, this wires EngraphisCloudDecisionClient into the route planner; that planner submits a purpose="custom" request, causing the cloud client to refresh credentials and make a network call. This contradicts the canonical hosted contract that managed Jev does not perform route selection and that managed custom/query_planning requests fail locally before credential refresh, so the advertised managed configuration attempts an unsupported operation and then degrades to fallback. Keep route planning BYOK-only or reject the managed path before constructing the remote planner.
AGENTS.md reference: AGENTS.md:L281-L288
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: c76873dc06
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| PUBLIC_OFFLINE_ARTIFACT = "offline-fixtures-v128.json" | ||
| PUBLIC_OFFLINE_SHA = "73f2d1a8cd6e2db070577582a2266f6605efc052800da17db9938f7a274bf755" | ||
| PUBLIC_OFFLINE_ARTIFACT = "offline-fixtures-v149.json" | ||
| PUBLIC_OFFLINE_SHA = "d5d36c55c4303d77b161137521dd31f77f39b7a0c9e2fed3ddb63e303b12cc6d" |
There was a problem hiding this comment.
Fresh evidence beyond the earlier v148 finding shows that the newly pinned v149 artifact itself hashes to 23b5d2bf01f2…, while this constant, its .sha256 sidecar, BENCHMARKS.md, and the generated visuals all claim d5d36c55c430…. Consequently, python -m pytest tests/ -q -x fails in test_public_numeric_evidence_registry_is_complete_and_live, blocking the required offline gate; regenerate or repin the v149 artifact and update every checksum consumer.
AGENTS.md reference: AGENTS.md:L50-L54
Useful? React with 👍 / 👎.
This branch has not been deployed
1 participant
Summary
Adds opt-in Jev evidence review and bounded route selection while keeping local deterministic behavior as the default. A caller-supplied planner that lacks advisory support now keeps its normal multi-route plan, and compact grounded retrieval previews use the same field allowlist as compact recall so exact values and private metadata are omitted.
The published managed-access contract covers eligible Pro users and eligible Team named seats, including paid viewers with active paid, trial, or test entitlements. It documents individual rolling limits and the existing production service-cap conflict. Managed availability remains disabled.
Managed Jev contract
Managed Jev remains
not_yet_availablepending release acceptance and capacity qualification. Once available, each eligible individual receives 100 evaluated questions per rolling hour, 1,000 per rolling five hours, and 2,000 per rolling 24 hours. The limits are per person, independent, and not pooled or monthly. Every evaluated question counts; a batch cannot bypass the limits, and command review uses two questions. There is no additional customer charge or customer provider key.The existing production fleet guard is 100 questions per day, which conflicts with these individual limits and must be resolved before launch. Cloud PR #102 is currently at
bc254d64, which pools Team usage and omits the five-hour limit; that head is not merge-qualified. No quality or cost-saving benefit is claimed.Evaluation
The opt-in BYOK planner probe on 40 public synthetic tasks selected a route on all 40 calls but left nDCG@5, Recall@5, and answer-token coverage unchanged. It does not establish improvement on held-out workloads.
Verification
fd5b060adb9cb90aa04f856c69f5628f309c1b6b.