kmp-memory

kmp-memory is a skill for Claude Code, Codex from underpass-ai/kmp. It costs 124 tokens per session (7,740 once invoked), scanned A, original, Apache-2.0.

A guide for using KMP as graph-based, time-aware memory for coding agents. It explains how to recover context, ask questions, move through history, check evidence, and record decisions.

In plain words
What is it for?
Use it when continuing earlier work, asking about stored project evidence, checking what changed over time, tracing claims to proof, or recording decisions.
Why use it?
It helps an agent choose the memory action that matches the user's request and return UNKNOWN when stored evidence does not support an answer.

Skill for Claude CodeCodex

Part of the kmp plugin — 11 skills, 10 commands, 2 hooks, 1 MCP server shipped together

Install

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.

agentmods
npx agentmods add skills/underpass-ai/kmp/kmp-memory
Any agent
npx skills add underpass-ai/kmp --skill kmp-memory
Clone the repo
git clone --depth 1 https://github.com/underpass-ai/kmp

Made for: Claude Code, Codex.

Or install kmp, the plugin that ships this one along with the rest of its 11 skills, 10 commands, 2 hooks, 1 MCP server.

Wrote 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.

agentmods badge for kmp-memory

README.md
[![agentmods](https://agentmods.dev/badge/skills/underpass-ai/kmp/kmp-memory.svg)](https://agentmods.dev/skills/underpass-ai/kmp/kmp-memory)
Your own site
<a href="https://agentmods.dev/skills/underpass-ai/kmp/kmp-memory"><img src="https://agentmods.dev/badge/skills/underpass-ai/kmp/kmp-memory.svg" alt="Measured on agentmods" height="20"></a>
Per session 124 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 7,740 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00124 $0.07740
Opus 5 $0.00062 $0.03870
Sonnet 5 $0.00025 $0.01548
Haiku 4.5 $0.00012 $0.00774

Measured today against content hash 993c43360b8c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

kmp-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 today.

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.

plugins/kmp/skills/kmp-memory/SKILL.md · 650 lines

How it starts

The opening of the file, as written. The whole thing — 650 lines — stays where its author put it; the contents beside it link to each section on GitHub.

KMP agent memory

KMP is graph-temporal memory for agents, reachable over MCP as ten memory tools plus three semantic view tools. It is a kernel, not a model: every answer is derived from stored evidence by construction. Nothing here generates prose. If the memory does not support an answer, kmp_ask returns UNKNOWN — that is a correct result, not a failure to work around.

Invoked, not assumed

KMP is opt-in. This skill governs what happens once something selects it; it does not claim every session. Four things select it:

  • the user names KMP or its memory, in any language — "usa kmp", "what does memory say", "check the store";
  • a /kmp:* skill or command runs;
  • the project's own instructions (CLAUDE.md, AGENTS.md) opt in;
  • the MCP initialize instructions report always-on routing, which an operator turns on deliberately with kmp-mcp config memory-routing always.

Without one of those, do the work from the material in front of you and make no KMP call. An unbidden kmp_wake against an empty or unrelated store is not a free no-op: it spends a round trip and can shape the answer with evidence nobody asked for.

Everything below is about a route already underway. Temporal precedence, page continuation, do not leave for repository files mid-page — those govern a KMP call in flight. None of them is a reason to start one.

Use this as a router, not a tool glossary

Choose a lane before the first call, then let every result choose the next move. Do not select kmp_ask once and keep treating the whole task as semantic when the evidence says it is not.

Once invoked, known work enters through kmp_wake; apply the remaining rows to the part of the goal the wake packet did not already answer.

Signal in the user's goal First move
Continue known work or recover its state kmp_wake
Yesterday, since, before/after, a date, what changed, current/latest/recent state, why now, or a release/decision window kmp_goto, kmp_near, kmp_rewind or kmp_forward
A genuinely non-temporal question answerable from stored evidence kmp_ask
One cited ref must support a consequential claim kmp_inspect
A connection between two refs is part of the claim kmp_trace
The user asks to see, show, open, or navigate memory — including muéstrame, enséñame, abre or ver Finish the retrieval lane, then kmp_view_open and kmp_view_apply_intent
A durable decision, constraint or outcome was reached kmp_write_memory

Read the full file on GitHub · 650 lines

Changes

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.

  1. today Changed · +58 lines 993c43360b8c
  2. yesterday Changed · +22 lines · +5 tokens per session 666719bd39a4
  3. 4d ago First seen · 570 lines · 119 tokens per session scan A 075c913abb89

Subscribe to this mod's changes

kmp-memory is a skill published in the GitHub repository underpass-ai/kmp (0 stars, last pushed today), licensed Apache-2.0. It adds 124 tokens to every session and 7,740 once invoked, about $0.0006 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 2026-08-31.

Related

Other skills, from other repositories

media-ingest

Ingest video, audio, PDF, book, screenshot, and GitHub repo content into the brain. Multi-format handling with entity extraction and backlink propagation. Covers video-ingest, youtube-ingest, and book-ingest subtypes.

garrytan/gbrain · 52 tokens

mem0-oss-to-platform

Plan and then execute a migration of a project from the mem0 open-source / self-hosted SDK (the local Memory class) to the mem0 Platform / hosted / managed SDK (the MemoryClient class). Use this whenever a developer wants to move, switch, or migrate their mem0 usage off OSS/self-hosted to the hosted API — e.g.…

mem0ai/mem0 · 273 tokens

Cortex

Operate Cortex, the LifeOS memory system — the typed Knowledge Archive (People, Companies, Ideas, Research with typed related: links) plus recall of prior work sessions, ISAs, and conversations. Search, add, harvest, develop, ingest, distill, graph-navigate, recall. USE WHEN cortex, knowledge, knowledge base, search…

danielmiessler/LifeOS · 196 tokens

agent-memory

../../../engineering/agent-memory/skills/agent-memory/SKILL.md.

alirezarezvani/claude-skills · 0 tokens

memory

Use when the user asks to remember, recall, forget, update, search, or inspect durable OpenSquilla memory, including profile facts in USER.md and long-term notes in MEMORY.md or memory//.md.

opensquilla/opensquilla · 44 tokens

ha-data-stores

Map of Hope Agent's local data stores and safe read-only query workflow. Use when the user asks where Hope Agent stores data, wants to inspect sessions/messages/memory/logs/background jobs/knowledge indexes/settings, asks the model to query local app data, or debugging requires checking persisted state. Trigger…

shiwenwen/hope-agent · 115 tokens