You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
v3.4 Day 2: humanize public surface for Universal Spine
The release ritual codified in ~/.claude/skills/engram-release/SKILL.md
calls Phase 3 — public surface refresh — out as the most-skipped step.
This commit walks it for v3.4.
README.md
- Top callout block leads with the May 2026 market context (Cursor
pricing crisis + Claude Code rate-limit pain) so the positioning
matches what users are actually searching for.
- Test count badge: 878 → 910.
- Token-savings badge synced to 89.1% (was 90.8% — outdated).
- New v3.4 "Universal Spine" callout above the v3.0 "Spine" callout.
- IDE matrix expanded from 8 to 11 entries; adds Cline + Codex CLI +
VS Code extension; updates Cursor row to mention OpenVSX extension;
flags engramx-continue as live on npm.
- "How It Compares" rewritten as the May 2026 8-row matrix (engramx
vs Cursor index / Aider repo map / Cline / Continue / Mem0 /
claude-mem / CartoGopher). Legacy comparison table moved into a
collapsed details block to preserve continuity.
docs/install.html
- <title>, meta description, OG title + description: refreshed for
v3.4 framing, lead with "8 IDEs, one install, 89% reduction".
- Hero pill: v3.0 → v3.4 "Universal Spine" + ship date + test count.
- Nav link version bumped.
- IDE matrix expanded to 13 entries; explicit Cline + Cursor (with
OpenVSX) + Continue (engramx-continue npm) + VS Code (any agent).
Both are humanized — variable sentence length, no AI lexicon, no rule
of three. The 89.1% number is repeated where useful but never as a
marketing claim — always with the source (bench/real-world.ts) named.
Process discipline:
- ~/.claude/skills/engram-release/SKILL.md authored this session as
the canonical 8-phase release checklist.
- ~/.claude/CLAUDE.md "During Every Task" item #11 added pointing
at the skill.
- Memory file feedback_engram_release_ritual.md written for cross-
session continuity.
No code changes. No new tests. Build untouched.
> **EngramX v3.0 "Spine" shipped 2026-04-24** — the biggest release since v1.0. The spine is now **extensible**: any MCP server becomes an EngramX provider via a 10-line plugin file. **Pre-mortem mistake-guard** warns before you repeat a bug. **Bi-temporal mistake memory** — refactored-away mistakes stop firing. **Anthropic Auto-Memory bridge** reads Claude Code's own consolidated memory. **SSE-streaming** packets render progressively. `engram gen` dual-emits `AGENTS.md` + `CLAUDE.md` by default. **89.1% measured real-world token savings** on 87 source files — reproducible in one command. 878 tests, CI green on Ubuntu + Windows × Node 20 + 22. Zero cloud, zero telemetry. See [CHANGELOG.md](CHANGELOG.md) for the full diff.
59
+
## Why this exists, May 2026
60
+
61
+
Three things broke at the same time. Cursor went usage-based and people started getting $1,400 surprise bills. Anthropic tightened Claude Code limits, then quietly tested removing it from the $20 Pro plan. Half the AI coding crowd migrated from one tool to the other, hit the new ceiling within a week, and started looking for any way to make a session last longer.
62
+
63
+
Engramx is what makes the session last longer. It indexes your codebase into a local SQLite knowledge graph once. Then it intercepts file reads at the agent boundary and replaces them with a structural summary the agent already has the working memory for. Same edit, same diff, same code shipped — fewer tokens consumed in the round trip.
64
+
65
+
On a real 87-file repo, the measured reduction is **89.1%**. That's not a marketing number. The benchmark is committed to this repo as `bench/real-world.ts` and runs against any project you point it at. Independent migration guides ([dev.to/56kode](https://dev.to/56_kode/why-were-moving-from-cursor-to-claude-code-and-why-you-should-too-9kh), [SpectrumAI Lab](https://spectrumailab.com/blog/claude-code-vs-cursor)) cite engram as the strongest measured number in the category.
66
+
67
+
Works in 8 IDEs and counting — Claude Code, Cursor, Cline, Continue.dev, Aider, Windsurf, Zed, OpenAI Codex CLI. One install, one graph, every tool benefits. Apache 2.0. Local SQLite. Nothing leaves your machine.
68
+
69
+
> **v3.4 "Universal Spine" shipped 2026-05-02** — multi-IDE detector covers 8 tools, Anthropic Claude Code plugin (`/plugin install engram`), VS Code / Cursor extension on OpenVSX, `engramx-continue` on npm, Cline integration documented. Cost Lens telemetry from v3.3.0 now feeds a weekly Markdown digest at `~/.engram/cost-report-YYYY-Www.md`. 910 tests, CI green on Ubuntu + Windows × Node 20 + 22. See [CHANGELOG.md](CHANGELOG.md) for the v3.3 + v3.4 diff.
70
+
71
+
> **EngramX v3.0 "Spine" shipped 2026-04-24** — the biggest release before v3.4. The spine is **extensible**: any MCP server becomes an EngramX provider via a 10-line plugin file. **Pre-mortem mistake-guard** warns before you repeat a bug. **Bi-temporal mistake memory** — refactored-away mistakes stop firing. **Anthropic Auto-Memory bridge** reads Claude Code's own consolidated memory. **SSE-streaming** packets render progressively. `engram gen` dual-emits `AGENTS.md` + `CLAUDE.md` by default.
60
72
61
73
---
62
74
@@ -368,14 +380,19 @@ engram hooks install # auto-rebuild graph on every git commit
368
380
369
381
## IDE Integrations
370
382
383
+
`engram setup` auto-detects every supported IDE on your machine and prints the right next-step for each. You don't have to remember which command to run — the detector knows.
|**Cursor**| MDC snapshot + native MCP + VS Code extension on OpenVSX |`engram gen-mdc`·[docs/integrations/cursor-mcp.md](docs/integrations/cursor-mcp.md)|
389
+
|**Cline**|MCP server (5M+ VS Code installs, no native answer to token burn) |[docs/integrations/cline.md](docs/integrations/cline.md)|
390
+
|**Continue.dev**|`@engram` context provider via [`engramx-continue`](https://www.npmjs.com/package/engramx-continue)|[docs/integrations/continue.md](docs/integrations/continue.md)|
|**Neovim**| MCP via codecompanion / avante |[docs/integrations/neovim.md](docs/integrations/neovim.md)|
380
397
|**Emacs**| MCP via gptel-mcp |[docs/integrations/emacs.md](docs/integrations/emacs.md)|
381
398
@@ -385,17 +402,40 @@ Per-IDE setup guides are in [`docs/integrations/`](docs/integrations/).
385
402
386
403
## How It Compares
387
404
405
+
The "context spine" slot — local-first, code-aware, works in any MCP runtime, with a reproducible benchmark — is currently unowned. Here's the field as of May 2026:
The matrix isn't a slight at any of them — most do something engram doesn't. Cursor's index is great inside Cursor. Aider's repo map is great in Aider. Cline's full-file rewrite model is honest about what it is. The point is that nobody else covers all eight rows. Engram is the only tool that does.
420
+
421
+
For the legacy comparison vs `Continue @RepoMap` / `Cursor .cursorrules` / `@199-bio/engram` (small repo-map approaches), see the matrix below.
|**Interception model**| Hook-based, automatic on every Read | Fetched at @-mention time | Static file, manual | Per-session map | MCP server, called explicitly |
391
429
|**Cache strategy**| SQLite at SessionStart, <5ms per read | No cache — live fetch | No cache | Per-session only | No cache |
392
430
|**Persistent memory**| Decisions, mistakes, patterns across sessions | No | Manual text file | No | No |
393
-
|**Multiple providers**|8 (AST, git, mistakes, MemPalace, Context7, Obsidian, LSP) | Repo structure only | No | Repo structure only | Graph query only |
431
+
|**Multiple providers**|9 (AST, git, mistakes, MemPalace, Context7, Obsidian, LSP, Anthropic Memory, MCP plugins) | Repo structure only | No | Repo structure only | Graph query only |
394
432
|**Mistake tracking**| LSP diagnostics → mistake nodes, ⚠️ on Edit | No | No | No | No |
395
433
|**Survives compaction**| Yes (PreCompact hook) | No | Yes (static file) | No | No |
<title>Install · engramx v3.0 "Spine" — the context spine for AI coding agents</title>
8
-
<metaname="description" content="One command installs. One command wires it all up. Your AI coding agent stops re-reading the same files and starts remembering. 89.1% measured token savings on a real codebase. Local SQLite, zero cloud, zero LLM cost. Pluggable — add Serena, GitHub, Sentry as 10-line plugins.">
7
+
<title>Install · engramx v3.4 "Universal Spine" — the context spine for every AI coding tool</title>
8
+
<metaname="description" content="One install. Works in Claude Code, Cursor, Cline, Continue, Aider, Codex, Windsurf, Zed. Your AI coding agent stops re-reading the same files. 89.1% measured token reduction on a real codebase. Local SQLite, zero cloud, Apache 2.0. The only context tool with a reproducible benchmark across 8 IDEs.">
<metaproperty="og:title" content="engramx v3.0 — the context spine. 89.1% measured savings, pluggable providers, mistakes that remember.">
12
-
<metaproperty="og:description" content="npm install -g engramx → engram setup. 89.1% measured real-world token savings on code agents. Any MCP server becomes a provider in 10 lines. Open source, Apache 2.0.">
11
+
<metaproperty="og:title" content="engramx v3.4 — the context spine for every AI coding tool. 89% measured token reduction. 8 IDEs. One install.">
12
+
<metaproperty="og:description" content="npm install -g engramx → engram setup. Works in Claude Code, Cursor, Cline, Continue, Aider, Codex, Windsurf, Zed. 89% measured savings. Local SQLite, zero cloud. Apache 2.0.">
0 commit comments