I accidentally hit SOTA on agentic memory by using AI companions

context and memory layer for AI coding agents

Built to be the best context layer for AI coding agents.

graphCTX keeps repo knowledge close to the work: commands, conventions, decisions, and hard-won fixes. Developers spend less time re-explaining context and more time shipping, with local memory that is fast, private, and reliable.

$ curl -fsSL https://graph.coder.company/install | sh

See the benchmarks →

~1ms

p50 recall, flat at scale

100%

post-compaction solve

$0

no metering or API key

local

source stays on your machine

~/your-repo · AI coding agent

// without graphCTX

user: what's the deploy command? agent: I don't see one in the repo — you'll need to tell me.

// with graphCTX memory

user: what's the deploy command? agent: ./scripts/ship.sh --canary --wait mem:9f3a2c

Same agent, same prompt. graphCTX gives it the repo memory developers otherwise repeat.

[ 01 ] · how it works · architecture

How graphCTX manages memory.

The system is deliberately narrow: capture reliable coding facts, keep them valid as the repo changes, and return only the context the agent can use.

01 CAPTURE

Extract trusted coding facts

graphCTX reads package scripts, lockfiles, CI, editorconfig, AGENTS.md, and session episodes so the memory base starts from repo evidence, not model guesses.

02 ANCHOR

Bind memory to git state

Facts are valid against commits and branches. When code changes, memory moves with the DAG instead of drifting on wall-clock timestamps.

03 PROMOTE

Promote only durable knowledge

Session details can become workspace or user memory only after evidence gates. Secrets and low-trust prose stay out of durable context.

04 RANK

Choose the smallest useful set

A relevance gate scores topic drift, entities, and file scope so the agent gets specific context instead of a noisy memory dump.

05 DELIVER

Attach provenance to every recall

Returned memory is compact, budgeted, and tagged with source provenance, making it easier for developers to trust and audit what the agent uses.

[ 02 ] · benchmarks · measured, not claimed

Benchmarked against Supermemory.

Same coding-fact set, same queries. graphCTX runs locally and answers in ~1ms; a live Supermemory search round-trip measured ~494ms (p50), so recall stays fast and predictable during development. Reproducible: graphctx compare --deep.

METRIC

graphCTX (local)

Supermemory

delta

Ingest (p50)

0.71ms

network + async

local, instant

Retrieval (p50)

1.2ms

494ms

~410× faster

Retrieval (p95)

3.9ms

720ms

~180× faster

Recall accuracy

90%

—

direct-fact

Network

none

required

offline

Cost / month

$0

$19–$399

free

Latency stays flat at scale

Per-prompt retrieval p50/p95 as the workspace grows. Indexed lookup plus a bounded semantic re-rank keeps the hot path at ~1ms — a 5,000-fact monorepo retrieves as fast as an empty one.

10 facts p50 1.04ms · p95 1.37ms

510 facts p50 1.15ms · p95 3.94ms

5,010 facts p50 1.33ms · p95 22.27ms

Reliable after compaction

Post-compaction solve rate across 14 coding tasks. graphCTX restored the needed repo fact in every run.

A · no memory 0/14

B · recall API 3/14

C · graphCTX 14/14

// graphctx eval run --arms A,B,C

// scope: this compares local latency + cost on direct coding-fact retrieval for developer workflows. Supermemory targets general/conversational memory with cloud connectors, cross-document reasoning, and neural embeddings that graphCTX doesn't attempt.

[ 03 ] · install · three commands

Start using repo memory in 30 seconds

No account. No API key. No cloud setup. Install the CLI, connect your agent, and give every session repo-aware memory.

$ curl -fsSL https://graph.coder.company/install | sh

Prefer npm? npm i -g graphctx

01

$ curl -fsSL https://graph.coder.company/install | sh

install the CLI (detects Node / Bun)

02

$ graphctx install claude

wire your agent (claude · cursor · opencode · generic)

03

$ graphctx doctor

verify graphCTX is connected