engram CLAUDE.md

Project instructions for Engram, a local graph-based memory system for AI coding assistants. It stores project decisions, reasoning, and past problems as connected information that agents can inspect.

In plain words
What is it for?
Use it to search project memory before substantial work, follow the project’s write-and-review process, and keep decisions and lessons connected in the graph.
Why use it?
It helps an agent recover important project context and follow established decisions instead of repeating research or losing the reasons behind earlier choices.

Instructions file

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 instructions/techtheist/engram/claude-md
Clone the repo
git clone --depth 1 https://github.com/techtheist/engram
Per session 6,983 This file is loaded in full into every session.
When invoked 6,983 The same file — it is already loaded in full.
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 $0.06983 $0.06983
Opus 5 $0.03492 $0.03492
Sonnet 5 $0.01397 $0.01397
Haiku 4.5 $0.00698 $0.00698

Measured 3d ago against content hash 18b98b22cb29, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

engram CLAUDE.md 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 3d 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.

CLAUDE.md · 52 lines

How it starts

The opening of the file, as written. The whole thing — 52 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Engram — context for Claude Code

What this is: a graph-based, durable, inspectable long-term project memory for AI coding assistants (Claude Code first). Local-first, user-owned, graph-first UI. The reasoning/decision memory layer (why/decided/what-bit-us) — not a code-structure graph; the wedge is the editable IDE-embedded graph pane + conflict surfacing + local. The line since 0.8.0: the most powerful and feature-rich inspectable graph long-term memory for software development with AI agents, based on reproducible research.

THE GRAPH IS THE MEMORY OF RECORD — this file is only a keyword index. The session brief is auto-injected; search engram before any non-trivial work and follow the write-verdict protocol. Details, rationale, gotchas, and the full history live in this repo's graph (.engram/graph.tepin, daemon on 8787) and the eval workbench graph (eval/.engram, project eval, scientific ontology Claim/Method/Question/Finding/Source/Task). PLAN.md was retired 2026-08-04 (merged here); PLAN §… references in code comments point at git history.

Hard rules (locked — don't relitigate without reason)

  • Open-source, MIT; optimize for docs & DX. Product name "Engram Alpha", binary engram-alpha (repo/plugin namespaces stay engram; JetBrains package dev.techtheist.engram — don't rename).
  • Rust backend (rmcp, rusqlite bundled, sqlite-vec, fastembed, tepindb) — no Node runtime dep. Vue 3.5 + TS + Vite + Bun, Pinia, Tailwind 4, Vue Flow; bun run lint / lint:style in frontend/.
  • Embeddings/models local-only; no LLM in the daemon, ever (encoder-only mechanisms); models nominate, people judge.
  • Retrieval changes cite a measured eval/ run or they don't ship (since 0.8.0).
  • 8 node types / 7 sentence-shaped verbs on the default ontology; no new types, no relates_to; durability governs staleness; high-value edges are replaces/conflicts-with. Per-graph ontology/policy config exists since 0.7.0, but THIS repo's graph stays on the default ontology permanently.
  • Hard delete is user-only (pane); MCP deliberately has no delete/register/pin tools. Writes are silent; transparency is the pane. One workspace version for every crate, stamped from the tag; claude-plugin/.claude-plugin/plugin.json must match (test-enforced).
  • Multi-user & repo sync: out of scope permanently (future enterprise product). Dogfood on the aggressive skill variant (relaxed is the user default). Eval ladder max 1500 notes.

Read the full file on GitHub · 52 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. 3d ago First seen · 52 lines · 6,983 tokens per session scan A 18b98b22cb29

Subscribe to this mod's changes

engram CLAUDE.md is an instructions file published in the GitHub repository techtheist/engram (17 stars, last pushed 3d ago), licensed MIT. It adds 6,983 tokens to every session, about $0.0349 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.