Skip to content

Add typed hard/soft negative memory and information-preserving failure search - #16

Merged
TheDarkLightX merged 25 commits into
codex/fcis-shared-memory-pilotfrom
feature/typed-negative-memory-20261114
Nov 15, 2026
Merged

TheDarkLightX merged 25 commits into
codex/fcis-shared-memory-pilotfrom
feature/typed-negative-memory-20261114

Conversation

@TheDarkLightX

Copy link
Copy Markdown
Owner

Summary

Adds Gödel-inspired negative discovery memory to both Python stdio Research Kernel MCP and the hosted companion, preserving the distinction between hard and soft negative information.

  • Hard (hosted only): derived from locally produced, source-pinned bounded-check counterexample receipts; explicitly restricted to the exact checked recipe and original revision/grid. No family-level pruning follows from a single witness.
  • Soft (both): append-only, structured failed attempts capturing what was tried, why it failed, original assumptions and scope, reported witnesses, speculative generalization, alternative representations, and conditions for revisiting.
  • Maximally informative negativity: read APIs surface missing information, a reproducible next validation action, and deterministic information-completeness triage—not a truth score or assertion of optimal information gain.
  • Legacy negative evidence remains searchable as soft, corrected revisions retain old failures without losing current status, and soft notes flag transitive dependents for review without creating checker authority.
  • Hosted History inspector now distinguishes checked hard witnesses from advisory soft lessons.
  • Strict command validation, tests for authority forgery and historical negative retention, documented proof obligations for any widened/generalized pruning.
  • Functional-core source pin includes the new negative derivation module.

Interfaces

Hosted: rk_negative_record (new), rk_negative_memory (enhanced: insights + classification filter, legacy fields retained).
Python: rk_negative_record and rk_negative_memory (both new, only soft until an independent verifier exists).

Validation

Focused Python and hosted regression tests added; request CI for Python 3.11–3.13, hosted tests/typecheck/build.
Not deployed or merged by this PR. Existing source-version-pinned journals require their original functional-core version for replay.

Security / epistemic constraints

No requested publication or transfer of private research workspaces. No source assertions imported as trusted certificates. No claimed class-wide impossibility from one witness or a caller-declared theorem. See docs/NEGATIVE_MEMORY_CONTRACT.md.

@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Nov 15, 2026 •

Copy link
Copy Markdown
Codex Review Summary: ✅ Completed

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-11-15T02:06:46.196437Z e77efa9 Manual request
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" or "@codex security review".

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 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: 5c34e89806

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 internal/research_kernel_mcp/kernel.py Resolved

Copy link
Copy Markdown
Owner Author

@codex review

Follow-up review requested after fixing historical-search truncation (streaming all eligible records before top-k ranking), the signed 256-bit magnitude boundary, and replacement of pending invitations at capacity. Regression tests added for all three; checker and kernel source pins regenerated. Please review the current head for correctness and release blockers.

@TheDarkLightX
TheDarkLightX merged commit 9d541e9 into codex/fcis-shared-memory-pilot Nov 15, 2026
4 checks passed
@chatgpt-codex-connector

Copy link
Copy Markdown

Codex Review: Didn't find any major issues. 🎉

Reviewed commit: e77efa9dd4

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".

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.

1 participant