gcontext.ai
Build context like you build code
AI agents forget everything between sessions. Prompts duplicate. Context drifts. gcontext gives you deterministic context architecture that any agent can navigate.
$ curl -LsSf gcontext.ai/gcontext/install.sh | shWhy prompts don't scale
The problem
The shift
See it in action
Context modules give your agent structured files it can navigate by itself.
How context resolves
The agent navigates a file tree, not a vector database, not a prompt template. Deterministic, auditable, yours.
context/
integrations/postgres/
info.md: schema, tables, relationships
module.yaml: secrets: [PG_URL]
workflows/deploy/
info.md: deploy steps, environments
module.yaml: secrets: [SSH_KEY]
tasks/sprint/
info.md: current sprint goals
module.yaml: no secrets
How it works
$ gcontext initWhat happens without structure
Without context architecture, every session starts from scratch. With structured modules, the agent picks up exactly where it left off.
What the agent actually sees
Not folders. Deterministic context architecture. The agent gets exactly what it needs, nothing more.
Without gcontext
You are a helpful assistant. Our database is PostgreSQL on AWS RDS. The main tables are users, orders, products, sessions, analytics_events, notifications, payments, refunds... [800 more lines of context] The deploy process uses GitHub Actions. SSH into the server and run... Actually wait, we switched to Docker last month. The old SSH process is... [repeated across every conversation]
Prompt grows forever. Stale info. No structure.
With gcontext
context/
llms.txt ← nav index
integrations/
postgres/
info.md ← schema
module.yaml ← secrets: [PG_URL]
workflows/
deploy/
steps.md ← deploy steps
module.yaml
tasks/
weekly-report/
brief.md ← current goals
module.yamlAgent navigates to what it needs. Always current. Always scoped.
Describe your stack, gcontext structures it
Describe your stack the way you would to a new teammate. gcontext turns it into navigable context architecture.
Your conversations
Add information about Samperalabs. We build AI-powered developer tools for engineering teams...
Add an integration for Stripe, we handle webhooks for checkout events... Also add Linear, we use it to track all our issues...
Add a support workflow. When a ticket comes in, triage it by severity, pull the linked Linear issue, and draft a reply...
Add a task, TASK_123. A user reported a billing failure after upgrading, here's the charge_id, their email, and the error code...
Context architecture
// you define the context. it grows with you.
5 filesContext in, code out
The agent reads the module, resolves secrets, writes a script, and executes. Deterministic, no guessing.
read
// agent reads the integration module
generate
import psycopg, os
conn = psycopg.connect(os.environ["PG_URL"])
cur = conn.execute("""
SELECT plan, COUNT(*)
FROM users
WHERE created_at > now() - interval '7 days'
GROUP BY plan
""")
print(cur.fetchall())
// generates a script from what it learned
execute
$ uv run python -c '...'
plan | count
-----------+------
free | 31
pro | 12
enterprise | 4
47 users signed up this week.
// runs the script, returns the result
read
// agent reads the integration module
generate
import psycopg, os
conn = psycopg.connect(os.environ["PG_URL"])
cur = conn.execute("""
SELECT plan, COUNT(*)
FROM users
WHERE created_at > now() - interval '7 days'
GROUP BY plan
""")
print(cur.fetchall())
// generates a script from what it learned
execute
$ uv run python -c '...'
plan | count
-----------+------
free | 31
pro | 12
enterprise | 4
47 users signed up this week.
// runs the script, returns the result
maatapp.com
soundmurai.com
bleakai.com“It has transformed how I approach solving support tasks. From a boring task I was always procrastinating into something actually fun to do.”
Your workspace, your rules
gcontext is open source and runs locally. Your context lives in your repo as files, portable, yours to keep.
Open source
CLI
Create context modules, validate structure, sync with your repo. Runs in your terminal. Free forever.
$ curl -LsSf gcontext.ai/gcontext/install.sh | shHosted platform
Coming soongcontext Cloud
Everything the CLI does, plus a visual editor, built-in AI chat with your full context, and managed infrastructure.