Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/clawdbrunner/openclaw-graphiti-memory/skillnpx skills add clawdbrunner/openclaw-graphiti-memory --skill skillgit clone --depth 1 https://github.com/clawdbrunner/openclaw-graphiti-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/clawdbrunner/openclaw-graphiti-memory/skill)<a href="https://agentmods.dev/skills/clawdbrunner/openclaw-graphiti-memory/skill"><img src="https://agentmods.dev/badge/skills/clawdbrunner/openclaw-graphiti-memory/skill.svg" alt="Measured on agentmods" 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 | $0.00032 | $0.00475 |
| Opus 5 | $0.00016 | $0.00237 |
| Sonnet 5 | $0.00006 | $0.00095 |
| Haiku 4.5 | $0.00003 | $0.00047 |
Grade A, and why
hybrid-memory 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 5d 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.
What it actually says
Hybrid Memory System
We use two memory systems integrated into a single view:
- QMD (Vector Store): For retrieving documents, specs, and full content.
- Graphiti (Knowledge Graph): For retrieving facts, timelines, and relationships.
Primary Tool
For 95% of memory queries, use the hybrid search script. It queries both systems in parallel.
~/.openclaw/scripts/memory-hybrid-search.sh "your query"
Optional flags:
[group_id]— Specify agent group (default:openclaw-main)--json— Output JSON for programmatic use
Specific Tools (Advanced)
Only use these if the hybrid script fails or you need granular control.
Graphiti Only (Temporal/Facts)
Search for specific temporal facts:
~/.openclaw/scripts/graphiti-search.sh "your question" openclaw-main 10
Log new facts (IMPORTANT):
~/.openclaw/scripts/graphiti-log.sh openclaw-main user "Name" "Fact to remember"
QMD Only (Deep Document Search)
If you need more results or specific file filtering:
qmd search "query" -n 10
Recall Pattern
- User asks question ("What was the plan for the project?")
- Run Hybrid Search (
~/.openclaw/scripts/memory-hybrid-search.sh "plan for the project") - Synthesize Answer from both the temporal facts and document snippets found.
- If needed: Use
readto get the full content of a file found in the QMD results.
When to Use Which
| Question Type | Use |
|---|---|
| "What's in GOALS.md?" | Hybrid search → read file |
| "When did we discuss X?" | Hybrid search (Graphiti results) |
| "What did I say last Tuesday?" | Graphiti direct |
| "Find notes about architecture" | Hybrid search (QMD results) |
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.
- 5d ago First seen · 66 lines · 32 tokens per session scan A 975ef9f8515b
hybrid-memory is a skill published in the GitHub repository clawdbrunner/openclaw-graphiti-memory (104 stars, last pushed 1mo ago), licensed MIT. It adds 32 tokens to every session and 475 once invoked, about $0.0002 per session on Opus 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 2026-08-30.
Other skills, from other repositories
midmem-dev
Develop/iterate on the MidMem middleware core (this repo's packages/core — the shared "LLM Wiki" knowledge layer). Use when adding or changing MidMem capability — retrieval lanes, tiers/lifecycle, work-memory events, concept routing, claims, grounding, the MCP/CLI/hook surfaces, or its tests. Encodes the…
llm-wiki
Karpathy's LLM Wiki — build and maintain a persistent, interlinked markdown knowledge base. Ingest sources, query compiled knowledge, and lint for consistency.
rag-sync
End-of-session review and sync of learnings to the personal LightRAG knowledge graph.
rag-remember
Store a fact, decision, or observation into the personal LightRAG knowledge graph immediately.
pinecone-research
Agent RAG and long-term memory with Pinecone.
agent-v3-memory-specialist
Agent skill for v3-memory-specialist - invoke with $agent-v3-memory-specialist.