Curated marketplace review request: Research Registry plugin
What variant of Codex are you using?
App and CLI
What feature would you like to see?
Please consider Research Registry for the OpenAI curated marketplace.
Research Registry is a local-first, source-backed evidence ledger for Codex. It
adds read-only recall and explicit deposit workflows over a local STDIO MCP
server. Research is stored with immutable source versions, exact evidence
spans, caller-authored claims, typed relationships, review history, freshness,
and provenance.
Additional information
Source and merged v2 implementation:
- https://github.com/akovanda/Codex-research-skill
- https://github.com/akovanda/Codex-research-skill/commit/22bbeddbe4f22031810989e53cd601d0778bd9c0
Prepared curated-repository contribution branch:
Attempting to open a pull request against openai/plugins was rejected because
this account does not have permission to create PRs there.
Validation completed:
- OpenAI plugin manifest validation
- both bundled Skill validations
- marketplace and MCP JSON validation
- pinned source artifact installation in isolated Python 3.12
- installed CLI entry-point smoke
- upstream full SQLite and PostgreSQL 16 regression, packaging, local-installer,
shared-Compose, migration/backfill, backup/restore, and release-gate CI
Safety and privacy:
- personal local SQLite and content-addressed blob storage by default
- no Docker requirement for the personal path
- deposit is explicit-only and never publishes or approves material
- new deposits are private, unreviewed, and low-trust
- stored content is treated as untrusted data, never instructions
- authenticated shared mode enforces user/organization namespace isolation
- immutable runtime source pin; no moving branch or automatic release
- Apache-2.0 licensed
The plugin is packaged as SemVer 0.2.0-alpha.1; the Python distribution uses
the equivalent PEP 440 version 0.2.0a1.