The manual configuration is documented in the repository; this catalog review did not execute it.
Evidence record · captured 2026-10-09
Retenta Memory for Codex
An official Retenta skill for giving scheduled and multi-run Codex workflows a durable memory spine through a Retenta MCP capsule.
Editorial assessment
This is a source-only assessment. It does not claim runtime or security testing.
What it does
The Codex variant applies the same recall, progress-ledger, and durable-finding protocol while adding Codex-specific trigger guidance and an explicit exclusion for one-off questions.
Best for
- Scheduled Codex jobs and recurring automations
- Long tasks that continue across separate Codex sessions
- Multiple agents that need a shared, ordered state spine
Strengths in the source
- States both positive triggers and a clear non-trigger for work that does not need persisted context.
- Treats progress entries and synthesized memory as distinct records with different jobs.
- Documents how to stop cleanly when write quotas pause while reads remain available.
Limitations
- It requires manual local installation plus a separately configured Retenta MCP server.
- The repository warns that its bundled plugin HTTP MCP form was not smoke-tested; the manual configuration is the documented route.
- The source review does not establish runtime or security behavior.
When not to use it
- One-off work that is fully contained in the current context
- Environments where an external MCP endpoint or persistent secret cannot be configured
Reviewed by Retenta editorial on 2026-10-09; scope: source only.
Files included in this review
plugins/codex/skills/retenta-memory/SKILL.mdplugins/codex/README.mdpackage.json
Evidence and boundaries
No independent security scan or behavioral sandbox result is claimed for this catalog record.
- Public source snapshot
- plugins/codex/skills/retenta-memory/SKILL.md
- Revision
- c43c0fde6db395db5f4dff4027e56a4d30fba053
- License evidence
- UNLICENSEDThe pinned root package.json declares UNLICENSED and the skill directory has no separate license file.
- Last source change
- 2026-08-19
- Independent verification
- Published by Retenta, but not independently verified by a third-party reviewer.
- Upstream repository
- Not publicly accessible at capture; the immutable public snapshot above is the cited evidence.
Compatibility
Codex
documentedThe repository includes a Codex-format SKILL.md, a manual MCP setup recipe, and a plugin manifest.
Requirements and access
- A Retenta capsule MCP URL
- A named Retenta owner access key exposed through RETENTA_KEY
- Network access to the configured Retenta MCP endpoint
Referenced tools
recalllog_progressremembersearchasklist_sourcesforgetArchitecture before installation
Understand what the skill does—and what memory must do.
AI agent memory across runs: a practical architecture
A skill gives an agent a procedure. Durable memory preserves what happened, what remains open, and what the agent learned after the current context disappears.
Install from the source
The repository identifies the manual MCP and skill setup as the currently documented path.
View the SKILL.md at its source Verify the publisher's install documentationDocumented manual install
codex mcp add retenta --url <your Retenta MCP URL> --bearer-token-env-var RETENTA_KEY
mkdir -p ~/.agents/skills/retenta-memory
cp plugins/codex/skills/retenta-memory/SKILL.md ~/.agents/skills/retenta-memory/SKILL.mdThyme RAG Capsules
Make this skill’s state durable, governed, and portable.
A Retenta capsule lets agents recall prior work, append provenance-stamped progress, and carry the memory to the next run.