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 skills add yantrikos/yantrikdb-mcp --skill persistent-memorygit clone --depth 1 https://github.com/yantrikos/yantrikdb-mcpWrote 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/yantrikos/yantrikdb-mcp/persistent-memory)<a href="https://agentmods.dev/skills/yantrikos/yantrikdb-mcp/persistent-memory"><img src="https://agentmods.dev/badge/skills/yantrikos/yantrikdb-mcp/persistent-memory/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/yantrikos/yantrikdb-mcp/persistent-memory"><img src="https://agentmods.dev/badge/skills/yantrikos/yantrikdb-mcp/persistent-memory.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00084 | $0.01190 |
| Opus 5 | $0.00042 | $0.00595 |
| Sonnet 5 | $0.00017 | $0.00238 |
| Haiku 4.5 | $0.00008 | $0.00119 |
Grade A, and why
persistent-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 9d 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 — 101 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Persistent memory (YantrikDB)
You have a persistent, semantic, typed memory that survives across sessions,
provided by the yantrikdb MCP server. It is not a notes file: it does
similarity recall, tracks belief revisions, detects contradictions, learns
procedures, and answers "what did I believe at time t". Treat it as your
long-term memory, not as optional tooling.
Tool names may carry a harness prefix (mcp_yantrikdb_recall,
mcp__yantrikdb__recall, or bare recall). The unprefixed names are used
below; map them to whatever your harness registered.
Session start — one call
Call session with action="digest" once. It returns a single briefing:
narrative head, open decisions, unresolved conflicts, pending triggers, and
stale high-importance memories. Do this before answering anything
substantive; a session that starts blank when 500 memories exist is a bug,
and it is this one.
Before acting — recall, the right way
- Query with one short natural-language sentence (5–10 words), not a keyword list. Separate focused calls beat one broad one.
- For "what is the CURRENT value of X" (latest config, current owner, present
status), use
memorywithaction="chain_head"— similarity search returns the most-similar revision, which for values that change over time is often stale.chain_headreturns the actual current value. - Trust signals: hits whose
why_retrievedsays "aged", "superseded", or "rarely confirmed" are weak evidence. Prefer fresher hits, and say so if you act on a flagged one. - Before starting a task you may have done before, call
procedurewithaction="surface"— a learned method beats re-deriving one.
During work — capture as you go
- New durable fact →
remember. Be specific and searchable ("Client Foo approved the March pricing tier", not "they said yes"). Setimportance: 0.8–1.0 decisions and hard commitments, 0.5–0.7 useful context, 0.3–0.5 background. Setdomainandsource. - A stored fact changed →
correctwith areason— never a secondremember. This preserves history and prevents the contradiction the next session would otherwise inherit. - Entity relationship learned →
graphwithaction="relate". - A working method discovered (a scraping recipe that got past a blocker, a
benchmark invocation that finally worked) →
procedurewithaction="learn". - Do NOT store: secrets or credentials, ephemeral task chatter, anything derivable from the repo or files you already have. Memory is for what would otherwise be lost when this session ends.
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.
- 9d ago First seen · 101 lines · 84 tokens per session scan A c3c782923e39
persistent-memory is a skill published in the GitHub repository yantrikos/yantrikdb-mcp (24 stars, last pushed yesterday), licensed MIT. It adds 84 tokens to every session and 1,190 once invoked, about $0.0004 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
honcho
Configure and troubleshoot Honcho memory for Hermes.
qmd
Hybrid local search over notes, docs, and transcripts.
hermes-agent
Use, configure, theme, extend, and orchestrate Hermes Agent.
caveman-compress
Compress a memory file such as CLAUDE.md or a todo list into caveman format to save input tokens, keeping a readable backup. Trigger: /caveman-compress.
plur-create-engrams
Create or improve PLUR engrams from conversations, documents, decisions, observations, and explicit preferences. Use for memory extraction, engram authoring, or reviewing proposed memories, including global, scoped, pinned, retrieved, and provisional knowledge. Ordinary use of existing memories does not require this…
lemmalog
Externalize working memory and logical state into the lemmalog Datalog engine (MCP). Use for ANY multi-step task where state should outlive one context window or span agents: long investigations, debugging sessions, audits, multi-agent searches, systematic explorations, planning with many interdependent constraints…