Plug Claude Desktop, Claude Code, and Projects into one secure URL. Lockers, secrets, and workflow graphs — without shipping API keys to the model.
{
"mcpServers": {
"claude-locker": {
"url": "https://claudelocker.lovable.app/api/mcp",
"transport": "streamable-http",
"headers": { "Authorization": "Bearer lk_demo_<your-token>" }
}
}
}Every pack ships as a live, view-only demo running on sample data. Open the canvas, watch it work, then clone it under your own GitHub and keys.
12 packs · 8 ungated for testing
Multi-agent Claude pipeline that scans 13F filings for cluster insider activity.
Background Claude Code agent that opens PRs to clean up tech debt nightly.
Drop a brief, get email + ads + landing copy in one orchestrated flow.
Plug-and-play Claude artifacts for ops dashboards, KPIs and reviews.
Five small Claude agents that triage, draft, and schedule from your inbox.
Generates Vitest + Playwright tests for changed files in big monorepos.
End-to-end retrieval flow with chunker, embedder, reranker, and Claude answer node.
Three Claude agents classify, draft, and escalate Zendesk tickets in real time.
Monthly update artifacts: KPI grid, narrative, risk heatmap, ask block.
Reads schema diffs, writes safe SQL migrations, opens a PR with rollback notes.
Drop a meeting recording, get cleaned notes and synced Linear tickets.
Agents pull logs from AWS, GitHub, Okta and assemble audit-ready evidence packs.
Grant yourself full demo access — every locked stack opens in view-only canvas mode with sample data. Zero sign-up.
New primitives we're building into the platform. Each lab is a real, runnable surface — not a roadmap card.
Dynamic routing between Claude sub-agents based on task entropy.
Replay any MCP session deterministically against a new model.
Side-by-side diff of two prompt versions with token-level deltas.
Per-tool spend caps with auto-downgrade from Opus to Sonnet.
Run any stack against 200 golden tasks and score regressions.
Share read-only locker views across orgs without exposing values.
Visualise per-prompt token cost over a 7-day window.
Mirror traffic to a candidate model and diff responses offline.
Every recipe lists the exact secrets, MCP servers, full code, and one-line deploy command.
Wake up to a 5-bullet diff of every competitor page that changed overnight.
HMAC-verified webhook → Claude enrichment → HubSpot upsert with score.
Claude Code reviews every PR and posts a risk-banded comment with suggested tests.
Stripe + Postgres metrics → narrative artifact → Resend to your investor list.
Drop an .m4a, get a spec doc and a Linear issue with acceptance criteria.
Notion → pgvector → Claude with strict citation schema.
Every locker action is a typed function — call it from the dashboard, an agent, or your own backend.
rotateSecret(lockerId, name) lockerRotate a stored secret in place; downstream stacks pick up the new value on next call.
cloneStack(stackId, { toGithub }) stackFork a marketplace stack into your own GitHub org with your locker bound.
grantSeat(email, role) workspaceInvite a teammate with viewer, operator, or admin role on the workspace.
snapshotCanvas(flowId) stackPersist current canvas state (nodes, edges, viewport) as a versioned snapshot.
auditExport({ from, to }) workspaceStream a signed JSONL of every MCP call in the time window for SOC2 evidence.
bindProvider(locker, provider, key) lockerAttach a provider credential (OpenAI, Anthropic, Slack, GitHub) scoped to one locker.
scheduleRun(flowId, cron) stackRun any stack on a cron expression with optional jitter and per-run budget cap.
lockerDiff(a, b) workspaceDiff two lockers by secret names, rotation policies, and bound stacks.
revokeToken(tokenId) workspaceImmediately invalidate an MCP bearer token and purge cached sessions.
tailAudit({ tool }) workspaceLive-tail MCP audit events filtered by tool name, locker, or principal.
Each guide lives at its own top-level URL — indexed, cited, and linked to a deploy-ready recipe.