--- title: "mehmory" description: "Memory abstraction layer for Claude Code and Codex CLI: deterministic capture hooks plus model-driven skills over a markdown wiki in ~/.mehmory." canonical: https://agentpluginsdirectory.com/plugins/mehmory last-updated: 2026-09-22 --- # mehmory Memory abstraction layer for Claude Code and Codex CLI: deterministic capture hooks plus model-driven skills over a markdown wiki in ~/.mehmory. - Slug: mehmory - Publisher: Christopher Freddy Getsfred - Repository: https://github.com/elderfo/mehmory - Manifest: plugin.json - Version: 0.4.0 - License: MIT - Category (editorial): agent-tooling - Skills: 6 (integrate, lint, onboard-session, pause, remember, resume) - MCP servers: 0 - Stars: 0 - Repository created: 2026-07-29 - Repository last pushed: 2026-09-07 - Publisher type: User - Listing: https://agentpluginsdirectory.com/plugins/mehmory - Schema: https://agent-plugins.org/schemas/1.0.0/plugin.schema.json ## What mehmory does, in the publisher's words Hook-enforced, model-maintained markdown wiki memory for Claude Code and Codex CLI. A deliberately "meh"-tier improvement in harness memory and continuity, built from boring parts: your project's memory is a git-backed directory of markdown at ~/.mehmory that a Claude Code or Codex CLI session reads and writes with ordinary file operations. No embeddings, no MCP server, no external services. Website: https://elderfo.github.io/mehmory/: the pitch, the quickstart, and how it works. mehmory ships a portable Agent Plugins v1.0.0 package: the root plugin.json and skills/ directory are client-neutral. Claude Code is supported through its compatibility manifest in.claude-plugin/ and the bundled lifecycle hooks; Codex CLI uses the same skills plus its native hook installation path. - docs/CONFIG.md: every config group, its real default, and which keys are actually wired up versus present but currently inert, including the per-harness hosts.claude-code.enabled / hosts.codex.enabled toggle. From the project README, punctuation lightly normalized. Full text: https://raw.githubusercontent.com/elderfo/mehmory/HEAD/README.md ## Skills - integrate: Merge captured mehmory inbox entries into the wiki pages under ~/.mehmory: editing pages, links, index lines and frontmatter, then clearing the inbox transactionally and committing. Use when the SessionStart nudge says the inbox is over threshold, or whenever the user asks to integrate, process, o… - lint: Full-sweep health check of the mehmory wiki under ~/.mehmory: staleness, orphan pages, contradictions across pages, and archive candidates. Reports everything first and applies only the fixes the user approves. Use when the user asks to lint, audit, clean up, or check memory. Writes to ~/.mehmory… - onboard-session: Seed mehmory for the current project from scratch, survey the repo (README, package manifest, git log, docs), then write project.md, a first set of pages and the index under ~/.mehmory, and commit. Use on a fresh store, when SessionStart says the store was just created, or when the user asks to on… - pause: Stop mehmory capturing and injecting for the rest of this session, no inbox writes, no context injection, no pointers. Use when the user says pause memory, stop capturing, mute mehmory, or is about to work on something they do not want remembered. Writes one flag file under ~/.mehmory/.state (outs… - remember: Save a fact, decision, correction or gotcha to the mehmory inbox right now, so the next integrate files it into the wiki. Use when the user says remember this, save this, note that, or do not forget. For a one-liner there is a faster path with no skill load at all, start a prompt with the `remembe… - resume: Turn mehmory capture and injection back on for this session after /mehmory:pause. Use when the user says resume memory, unpause, or start capturing again. Clears one flag under ~/.mehmory/.state (outside the project), so Claude Code may prompt for permission. It never re-enables a hook that was swi… Descriptions come from the frontmatter of each SKILL.md, punctuation lightly normalized.