// product

Context from your entire stack (not just your codebase)

CoopAI indexes your entire codebase and the tools that your team uses. Providing context from tools like GitHub, GitLab, Slack, Jira, Notion, and more.

token_validator.tsauth_routes.go

CoopAI

1import { validateSession } from './auth';

3export async function validateSession(payload: JwtPayload) {

4 // Validates signature before route handlers

5 if (!payload?.signature || !payload?.exp) {

6 throw new AuthError('empty_or_unsigned_payload');

7 }

8 return validate(payload);

Plays prompt → context → outcome · advances when complete · hover to pause

// works with your models

Anthropic
OpenAI
Google
Fireworks.ai

// quick_actions

Quick actions & code creation

Code with your team's full context — one-click intelligence workflows plus inline complete and in-file edit.

coop understand-repo
Understand Repo
Architecture, ownership, and key files — without cloning the whole codebase.
coop trace-decision
Trace Decision
Why this code exists. Pull rationale from commits, PRs, and team context.
coop find-owner
Find Owner
Who owns this area and the escalation path when you need a human.
coop blast-radius
Blast Radius
Impact of changing this code — integrations, APIs, and operational risk.
coop knowledge-gaps
Knowledge Gaps
Missing context and blind spots before you ship.
coop complete
Inline complete
Ghost-text completions as you type — single- and multi-line, Tab to accept. Optional graph context (Pro) adds dependents and ownership from your indexed repo.
coop edit
Edit selection
Highlight a block, describe the change, review an inline diff. Accept, retry, or undo — craftsmanship in the editor, not an autonomous agent rewriting your tree.
coop complete --graph
Completion-only routing
Inline requests use a separate zero-retention code-completion path (`x-use-case: code-completion-only`) — distinct from chat, with keys on your server.
coop chat
Chat
Free-form questions with repo context, saved prompts, and streaming responses from your choice of model.

Who owns services/billing/invoice_handler.go? I need to change idempotency keys — CODEOWNERS says @platform-payments but git blame shows @marcus. Does pkg/ledger/posting.go share the same on-call rotation?

CODEOWNERS + blamecross-package owners

If I refactor TokenValidator.validate() in internal/auth/token_validator.ts, what breaks downstream? List dependents in api-gateway, workers/webhook-processor, and any shared libs that import this symbol.

symbol graph dependentscross-service impact

Who should review my PR touching migrations/008_session_index.sql, auth_middleware.go, and session/store.go? Want reviewers with context on incident #inc-auth-992 and recent changes to the OAuth refresh path.

blame-aware reviewersincident + PR context

I'm onboarding to coop-backend — where does webhook ingestion start, and how do events flow into the job queue vs GraphCache? What are the 5 files I should read first to trace a GitHub push end-to-end?

entrypoints + data flowread order by layer

Before I ship changes to GraphConsistencyManager.applyEvent(), what am I missing? Any Slack threads or Jira tickets on webhook dedupe, and who last modified the Slack normalization path in handlers/slackWebhookHandler.ts?

undocumented decisionsSlack + Jira cross-ref

Pull the Slack thread and Jira ticket tied to auth_middleware.go — why did we add zero-retention headers here? Cross-reference commits on internal/llm/router.go from the last 90 days and link the original design note.

#platform-auth threadPROJ-1847 + commits

I'm finishing the empty-payload guard in token_validator.ts — complete the if using the same AuthError pattern as the rest of billing/auth and what our graph shows for downstream callers.

ghost-text completegraph · AuthError pattern

Select the refresh-token branch in oauth_refresh.ts and edit it to match rejection semantics from token_validator.ts — throw AuthError('empty_or_unsigned_payload') instead of returning null.

⌥K edit selectioninline diff accept

CODEOWNERS + blamecross-package owners

Click to copyDemo

// lightning_mode

Lightning Mode: Instance-wide indexed search

For teams with many interconnected repositories, Lightning Mode provides instant cross-repo context through instance-wide indexing.

What it includes:

Search across your entire codebase instantly
Instance-wide symbol and content indexing
Instant context without repository cloning
Smart caching and incremental updates
Works offline (falls back to local index)

Who needs it:

Teams with 5+ connected services, microservice architectures, or teams where cloning monorepos isn't practical.

Available in: All plans (free: up to 3 Deep-Indexed repos org-wide; Pro: unlimited)

// write

Graph-grounded code creation

Stay in the file. Write like you've been in the repo for years.

Inline complete

Ghost-text completions as you type — single- and multi-line, Tab to accept. Optional graph context (Pro) adds dependents and ownership from your indexed repo.

Edit selection

Highlight a block, describe the change, review an inline diff. Accept, retry, or undo — craftsmanship in the editor, not an autonomous agent rewriting your tree.

Completion-only routing

Inline requests use a separate zero-retention code-completion path (`x-use-case: code-completion-only`) — distinct from chat, with keys on your server.

// features

Capabilities

// indexing

01
Remote knowledge graph
CoopAI indexes repositories via webhooks and background jobs. The extension queries ownership, dependents, and decision signals without requiring a full local clone.
02
Graceful degradation
When graph data is unavailable, CoopAI falls back transparently and tells you what context is missing instead of hallucinating.

// write

03
Inline complete & edit
Ghost-text completions and selection-based edits in the open file. Optional graph context (Pro) and project-style biasing keep suggestions aligned with your codebase — craftsmanship in the editor, not autonomous agents.
04
Completion-only routing
Inline requests use a dedicated zero-retention path (`x-use-case: code-completion-only`) — separate from chat, with keys on your server.

// team_context

05
Editor context menu
Right-click any selection to Trace Decision, Find Owner, Blast Radius, Understand Repo, or surface Knowledge Gaps.
06
Workspace prompt library
Save and share team prompts in `.coop/prompts.json`. Run common workflows from the sidebar or context menu with one click.
07
Slack & ticket context
CoopAI connects organizational context — Slack threads, tickets, and PR history — so answers reflect how decisions were actually made.

// platform

08
Multi-model chat
Stream responses from Anthropic, OpenAI, Gemini, and more. Provider keys live on your CoopAI server — never in the IDE or on developer laptops.

Inline autocomplete is opt-in — enable coopAI.autocomplete.enabled in settings. Autocomplete docs. Edit selection ships with apply and undo — see Edit mode docs.

coop — zsh

coop demo --repo yours

See CoopAI on your codebase

Book a demo with our team or install the free VS Code extension from the Marketplace.

Book a demo Install extension

code intelligence from your entire stack

Context from your entire stack (not just your codebase)

Quick actions & code creation

Lightning Mode: Instance-wide indexed search

What it includes:

Who needs it:

Graph-grounded code creation

Inline complete

Edit selection

Completion-only routing

Capabilities

Remote knowledge graph

Graceful degradation

Inline complete & edit

Completion-only routing

Editor context menu

Workspace prompt library

Slack & ticket context

Multi-model chat

See CoopAI on your codebase