Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add atman-33/workhub/plugin install memoryWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/atman-33/workhub/memory-setup)<a href="https://agentmods.dev/skills/atman-33/workhub/memory-setup"><img src="https://agentmods.dev/badge/skills/atman-33/workhub/memory-setup/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/atman-33/workhub/memory-setup"><img src="https://agentmods.dev/badge/skills/atman-33/workhub/memory-setup.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00070 | $0.00811 |
| Opus 5.5 | $0.00028 | $0.00324 |
| Sonnet 5.5 | $0.00014 | $0.00162 |
| Haiku 4.5 | $0.00007 | $0.00081 |
Grade A, and why
memory-setup scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 27d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 68 lines — stays where its author put it; the contents beside it link to each section on GitHub.
memory-setup — Enable long-term memory on this machine
The workhub plugin ships memory hooks that save each session's Q&A pairs into a vault-local SQLite database and inject relevant past conversations into new sessions (hybrid FTS5 + vector search, fully local, no LLM). The hooks stay silently disabled until this one-time machine setup has run.
Steps
-
Run the idempotent setup script (safe to re-run; fast no-op when already set up):
node "${CLAUDE_PLUGIN_ROOT}/engine/cli.mjs" setupThis installs npm dependencies (node-sqlite3-wasm, @huggingface/transformers) into
~/.workhub/memory-engine/, downloads the embedding model (~320 MB, one time) into~/.workhub/memory-engine/models/, copies the engine to~/.workhub/memory-engine/engine/(the stable path the vault's OpenCode memory plugin calls), creates<vault>/_ai/state/memory.db, makes sure the vault.gitignoreexcludes the database, and writes the.setup-versionmarker the hooks and the workhub app check.The first run takes several minutes, almost all of it the model download. Warn the user before starting if they seem to be in a hurry. No dependency compiles from source, so a C/C++ toolchain (Visual Studio Build Tools) is never required.
-
Verify:
node "${CLAUDE_PLUGIN_ROOT}/engine/cli.mjs" statusReport the summary (setup state, database path, memory counts) to the user. From the next session on, memory capture and injection are active in both Claude Code (plugin hooks) and OpenCode (the vault's
.opencode/plugins/memory-plugin.ts). Per-agent on/off switches live in the workhub app settings (memory_claude_code/memory_opencode).
Troubleshooting
- npm install fails: the dependency set is pure JavaScript/WebAssembly
plus prebuilt binaries, so a failure here is a network or registry problem,
not a missing compiler. Check
node --version(needs Node 20+) and retry withnode cli.mjs setup --force. If an older install left abetter-sqlite3build error behind, it is stale — setup no longer uses it. - "unable to open database file" on a database from an older engine: it is
still in WAL mode, which the WASM SQLite build cannot open. The engine
converts it automatically on the next run; if the message persists, check
that Node is 22.5+ (
node --version) sonode:sqliteis available for the conversion. - Model download fails: network issue — rerun step 1; the download resumes from the HF cache.
- Engine version changed after a plugin update: hooks fall back to silent no-op; rerun step 1 (the marker version mismatch triggers a reinstall automatically).
- Setup never runs inside hooks; if the user does not want long-term memory on this machine, simply do not run this skill (everything stays disabled) and dismiss the app's setup notice via its settings.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 27d ago Changed f22b1b2c82bb
- 28d ago Changed c5471fe8fdb1
- 1mo ago First seen · 68 lines · 70 tokens per session scan A 15988a6180b5
memory-setup is a skill published in the GitHub repository atman-33/workhub (3 stars, last pushed 2d ago), licensed MIT. It adds 70 tokens to every session and 811 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
llm-wiki
Explain and apply the foundational LLM Wiki architecture and knowledge-management strategy. Use for setup, architecture, schema, or organization decisions; operational ingest/query/lint work uses dedicated skills.
wiki-export
Export the Obsidian wiki graph to JSON, GraphML, Neo4j Cypher, Postgres/SQL, HTML, or OKF bundles. Use when transferring or visualizing wiki data in external tools; wiki-import handles the reverse direction.
llm-wiki
Build and maintain a persistent, interlinked Obsidian-compatible markdown wiki using Karpathy's LLM Wiki pattern. Extension-backed with auto-generated metadata, guardrails, and 14 custom tools (+3 opt-in agent-trajectory tools).
garden-tending
Tend the knowledge garden — diagnose vault health, condense project memos into project.md overviews, create/merge cross-project topics, fix broken links, archive superseded notes, extend trails. Trigger on "tend the garden", "condense", "update project overview", "check vault health", "where are we with X?", "what…
curator-practice
Operating philosophy for autonomous memex curator tending — orientation protocol, signal triage, bounded work units, initiative thresholds, logging conventions. Trigger on cron-launched runs with no user prompt, "use your judgment", "tend without instructions", "do a pass on the vault", "I'll be away — do what needs…
memo-writing
How to write effective session memos — format, frontmatter schema, observation extraction, topic-signal append. Trigger on /memex:save, "save this for later", "remember this", "save what we discussed", "document this session", "create a memo", or when the [memex] activity nudge appears in context. Do NOT trigger for…