witan-memory

A shared knowledge graph for storing and retrieving project facts, reusable patterns, decisions, and lessons across agents, sessions, teammates, and repositories.

In plain words
What is it for?
Use it to load repository context at the start of work, save solutions to non-obvious problems, and record hand-off information or decisions for future contributors.
Why use it?
It prevents useful discoveries from being trapped in one agent's private memory or repeatedly rediscovered. It also keeps durable team knowledge separate from committed project instructions.

Skill for Claude CodeCodex

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/mitodl/agent-kit/witan-memory
Any agent
npx skills add mitodl/agent-kit --skill witan-memory
Clone the repo
git clone --depth 1 https://github.com/mitodl/agent-kit

Made for: Claude Code, Codex.

Per session 112 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,284 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 $0.00112 $0.02284
Opus 5 $0.00056 $0.01142
Sonnet 5 $0.00022 $0.00457
Haiku 4.5 $0.00011 $0.00228

Measured yesterday against content hash 408337eace21, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

witan-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 yesterday.

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.

mcp/servers/witan/witan/skills/witan-memory/SKILL.md · 224 lines

How it starts

The opening of the file, as written. The whole thing — 224 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Memory

The team's shared knowledge graph stores four kinds of memories, all backed by Omnigraph and accessible via the witan MCP server. The repo is auto-detected from .git/config — you rarely need to pass it explicitly.

This graph vs. CLAUDE.md / AGENTS.md

CLAUDE.md / AGENTS.md hold static, human-committed project instructions that load into every session automatically. This graph holds facts, patterns, and lessons you accrue and query on demand, shared across agents and repos. Don't duplicate committed CLAUDE.md content into the graph; use the graph for what you learn while working that isn't already written down.

This graph vs. your built-in / session memory

Your coding agent also has a private, on-disk built-in/session memory (e.g. Claude Code's memory/ files). That store is local to one machine and one user; witan is shared across the team, synced, and queryable by every agent and repo. They overlap in purpose, so choose deliberately:

  • Store in witan anything another agent, a future session, or a teammate would benefit from: project facts, reusable patterns, lessons from a mistake, decisions and their rationale, hand-off context — especially when it's tied to a repo, code symbol, task, or project. This is the default for engineering knowledge worth keeping.
  • Keep in built-in/session memory only what is private or non-shareable: machine-local paths, personal scratch notes, harness/tooling preferences, or ephemeral state for the task in front of you right now.

Rule of thumb: if it's durable AND shareable, it goes in witan. When in doubt, prefer witan — a teammate finding your lesson is the entire point; a note stranded in one machine's session memory helps no one else. If you catch yourself about to write an engineering fact, pattern, or lesson to local session memory, store it here instead (or as well, if the harness mandates a local write).

When to Use Each Tool

recall — load context (the default read)

Read the full file on GitHub · 224 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. yesterday First seen · 224 lines · 112 tokens per session scan A 408337eace21

Subscribe to this mod's changes

witan-memory is a skill published in the GitHub repository mitodl/agent-kit (2 stars, last pushed 3d ago), licensed BSD-3-Clause. It adds 112 tokens to every session and 2,284 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

article-writing

Write articles, guides, blog posts, tutorials, newsletter issues, and other long-form content in a distinctive voice derived from supplied examples or brand guidance. Use when the user wants polished written content longer than a paragraph, especially when voice consistency, structure, and credibility matter.

affaan-m/ECC · 57 tokens

a-evolve

Apply A-Evolve's agentic evolution methodology to improve AI agent performance across runs. Use when the user wants to diagnose agent failures, generate targeted skills from error patterns, evolve system prompts, or accumulate episodic knowledge. Works standalone or inside AutoResearchClaw pipelines. Triggers on…

aiming-lab/AutoResearchClaw · 100 tokens

hive.chart-creation-foundations

Required reading whenever any chart tool is available. Teaches the one-tool embedding contract (call chartrender → live chart appears in chat AND a downloadable PNG lands in the queen session dir), the ECharts (data viz) vs Mermaid (structural diagrams) decision, the BI/financial-grade aesthetic baseline (no…

aden-hive/hive · 133 tokens

code-documenter

Use when adding docstrings, creating API documentation, or building documentation sites. Invoke for OpenAPI/Swagger specs, JSDoc, doc portals, tutorials, user guides.

zebbern/claude-code-guide · 39 tokens

deck-course-module

暖纸背景 + Playfair, 左侧学习目标常驻, 含 MCQ 自测页.

nexu-io/html-anything · 25 tokens

creating-skills

Guide for creating Claude Code skills following Anthropic's official best practices. Use when user wants to create a new skill, build a skill, write SKILL.md, update an existing skill, or needs skill creation guidelines. Provides structure, frontmatter fields, naming conventions, and new features like dynamic context…

redai-infra/Relax · 70 tokens