Skip to content

feat: harden memory recall and evaluation - #97

Merged
Coding-Dev-Tools merged 13 commits into
mainfrom
codex/harden-memory-recall-eval
Jul 31, 2026
Merged

feat: harden memory recall and evaluation#97
Coding-Dev-Tools merged 13 commits into
mainfrom
codex/harden-memory-recall-eval

Conversation

@Coding-Dev-Tools

Copy link
Copy Markdown
Owner

Summary

  • Harden scoped recall, context packing, sync, consolidation, and dashboard graph interactions.
  • Add adaptive candidate-depth controls across service, MCP, and HTTP surfaces.
  • Expand reproducible evaluation, privacy-safe evidence export, and context-savings accounting.

Validation

  • python -m pytest tests/ -q
  • python -m eval.harness --dataset eval/datasets/sample.jsonl --k 5
  • python -m eval.harness --dataset eval/datasets/codemem.jsonl --k 5
  • python -m eval.ablation
  • ruff check .

Comment thread eval/context_economy.py Fixed

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: f5e93cabb3

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".

Comment thread eval/benchmark.py Outdated
Comment thread eval/benchmark.py Outdated
Comment thread engraphis/dashboard_assets/engraphis-graph.js Outdated
Comment thread scripts/install_shortcuts.py
Comment thread eval/context_economy.py Fixed

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 826065b3ab

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".

Comment thread eval/benchmark.py Outdated
Comment thread eval/benchmark.py
Comment thread eval/agent_benchmarks.py
Comment thread eval/context_economy.py Fixed

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 891488c56c

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".

Comment thread engraphis/dashboard_assets/engraphis-graph.js Outdated

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 4e47f6f48e

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".

Comment thread eval/agent_benchmarks.py
Comment thread engraphis/dashboard_assets/engraphis-graph.js Outdated

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 8c7d03ff2f

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".

Comment thread eval/benchmark.py Outdated
Comment thread engraphis/classic_assets/dashboard.js

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 93fd0ee076

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".

Comment thread eval/benchmark.py

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: e9d4fd322e

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".

Comment thread engraphis/core/store.py
Comment thread eval/agent_benchmarks.py Outdated
@Coding-Dev-Tools
Coding-Dev-Tools merged commit 81e0b86 into main Jul 31, 2026
11 checks passed

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: e6a10f5bac

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".

Comment thread eval/benchmark.py
Comment on lines +256 to +259
try:
parsed = urlsplit(value)
except ValueError:
return value

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Fail closed when URL splitting rejects the authority

When a recorded command contains malformed IPv6 authority syntax with userinfo, such as https://user:password@[invalid/path, urlsplit() itself raises and this branch returns the original string, so report_envelope() publishes the credentials unchanged. Fresh evidence beyond the malformed-port fix is that this failure occurs before the new parsed.port guard; sanitize userinfo from the raw authority or redact the entire value on parse failure.

AGENTS.md reference: AGENTS.md:L206-L207

Useful? React with 👍 / 👎.

Comment on lines +172 to +173
if re.fullmatch(r"[0-9a-f]{40}", reader_revision) is None:
raise ValueError("reader_revision must be an immutable lowercase 40-character commit")

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Require a pinned evaluator revision

When --evaluator-model is supplied without a revision, or with an arbitrary mutable label, the evidence exporter accepts it even though the reported official QA scores may depend on that evaluator. The resulting immutable artifact cannot identify or reproduce the evaluator weights; require the evaluator model and a 40-character immutable revision together, just as this code already does for the reader.

Useful? React with 👍 / 👎.

@Coding-Dev-Tools
Coding-Dev-Tools deleted the codex/harden-memory-recall-eval branch July 31, 2026 05:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants