docs(blog): publish cross-harness memory CLI story

[TimeToBuildBob]

Publishes “One Memory CLI, Whichever Harness You Run” and its sharing card after the recall implementation merged and the same read-only command succeeded in real Claude Code and Codex sessions.

The article explains the deployed reader, root resolution, backend disclosure, and the limits of the retrospective retrieval benchmark. It separates the live Claude Code prompt hook from explicit Codex invocation and future integrations.

Validation: complete Jekyll/Tailwind build passed; scoped repository hooks and redacted content sync passed; independent review checked factual scope, the generated HTML, all code blocks and table rows, and the OG image. The benchmark table was reconciled against saved counts and the retrieval code; it uses equal rank cutoffs.

[TimeToBuildBob]

🤖 AI code review

Adds a blog post describing a cross-harness memory CLI for gptme, recounting a recall test performed in Claude Code and Codex and noting that the storage layer and lexical recall changes are merged. Includes commands and hook details for gptme-util memory operations, a benchmark table comparing the old retriever with the shared CLI TF-IDF retriever, and links to related GitHub issues and PRs.

Safe to merge — no P0/P1 findings

Confidence 5/5

✅ No findings. The diff looks correct to me on this pass.

_{Reviewed c3bd3337cf1c · openrouter/deepseek/deepseek-v4-flash-0731 · llm engine · 94s · about this reviewer}

_{Maintainer commands}

_{@TimeToBuildBob review (own line) — fresh review · @TimeToBuildBob fix — a worker acts on the findings. Once per comment; 👀 = received.}