Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/agentstacktech-agentstack-backend
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: backend, devops, ml, security
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/agentstack-backend
What's inside
- Skill name
agentstack-backend- Description
- Entry point for AgentStack backend ecosystem. Use WHENEVER the user mentions backend, API, server, database, auth, storage, hosting, publish site, static site, ZIP deploy, RBAC, payments, rules, webhooks, scheduler, RAG, subscriptions, messenger, integrations, or feature flags. Routes to domain-specific skills.
- License
- MIT
- Instructions
- 128 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# AgentStack Backend — Master Router
AgentStack is a **full backend ecosystem** exposed through ONE MCP tool: `agentstack.execute`. Capabilities come from live `GET /mcp/actions` (and `/agentstack-capability-matrix`) — do not hard-code action counts in skills.
## Quick router — pick the sub-skill
| User intent signals | Sub-skill | Primary MCP |
|---------------------|-----------|-------------|
| CRM / pipeline / contact | `agentstack-crm` | `crm.*` |
| business head / organ / command center | `agentstack-business` | `business.*` |
| AGNT / agUSD / AgentNet | `agentstack-agentnet` | `agentnet.*` |
| storefront studio / merchant | `agentstack-storefront-studio` | `commerce.storefront.*` |
| activate selling / first sale / seller onboarding | `agentstack-commerce` | `commerce.sell.activate` |
| project wallet / treasury | `agentstack-project-wallet` | `finance.project.*` + project wallet REST |
| professional services / hire studio /services | `agentstack-services` | REST `/api/public/services/*` (no MCP) |
| where in UI / next step | `agentstack-guidance` | `guidance.*`, discovery |
| store, data, config, 8DNA leaves | `agentstack-data` | `projects.patch_data`, `data_access.*` |
| sandbox / canary / generation fork / X-AgentStack-Env | `agentstack-data` | `generation.*` |
| publish site, /s/ URL, ZIP deploy | `agentstack-hosting` | `hosting.*` |
| support ticket, staff inbox, psup | `agentstack-support` | `social.support.*` |
| upload, quota, attachment, media | `agentstack-storage` | `storage.*`, REST upload |
| login, register, role, RBAC | `agentstack-auth-rbac` | `auth.*`, `rbac.*` |
| authorize plugin / Device Code / MCP missing in plugin | Connect on plugin MCP or `/agentstack-authorize` | Plugin `mcp.json` (OAuth) + Device Code → `~/.cursor/mcp.json` |
| when X then Y, automation, workflow | `agentstack-logic` | `logic.*`, `commands.*` |
| payment, wallet balance, checkout, buffs (not project treasury) | `agentstack-commerce` | `payments.*`, `wallets.*`, `buffs.*` |
| digital goods, asset wizard | `agentstack-commerce-assets` | `assets.*` |
| RAG, embedding, knowledge base | `agentstack-rag` | `rag.*`, `knowledge.policy_templates.*` |
| cron, webhook, notification; Stripe *callback* (not Checkout SDK) | `agentstack-signals` | `scheduler.*`, `webhooks.*` |
| project, API key, tenant | `agentstack-projects` | `projects.*`, `apikeys.*` |
| agent fleet, AI Builder | `agentstack-agents-ai` | `agents.*`, `ai_builder.*` |
| chat, DM, message ordering | `agentstack-messenger` | `social.*` |
| Slack, integration recipe | `agentstack-integrations` | `integrations.*` |
| where in UI, Compass, discover | `agentstack-discovery` | discovery manifest + UI registry |
| OpenAPI spec, REST surface, endpoint map | `agentstack-openapi` | OpenAPI + REST routing |
| hub task, capability atom | `agentstack-capability-tasks` | PTC manifests |
| TypeScript SDK, sdk.protocol | `agentstack-sdk` | `@agentstack/sdk` |
| Solana grant tooling (optional) | `solana-agentstack-mcp` | grant-scoped actions only |
Pick the **primary** bucket first; consult others by reference for multi-step flows.
**Disambiguation:** project treasury → `agentstack-project-wallet`; personal/commerce wallet → `agentstack-commerce`; inbound Stripe webhook → `agentstack-signals` / integrations (never `@stripe/stripe-js` for AgentStack checkout). Cursor plugin sign-in / “MCP not in the plugin” → `/agentstack-authorize`, not `auth.login`.
## User request → action (hot paths)
**Canonical SoT:** `docs/plugins/CONTEXT_FOR_AI_MCP.md` — edit the table there; run `node provided_plugins/scripts/sync-mcp-hot-path-skill.mjs`.
<!-- BEGIN:AUTOGEN-HOT-PATH-TABLE -->
| User request (example) | Domain | action(s) | Notes |
|------------------------|--------|-----------|--------|
| Create a project | Projects | `projects.create_project` (signed-in) or `projects.create_project_anonymous` (no key yet) | **Cursor plugin:** `/agentstack-authorize` or Connect — not anonymous. Anonymous returns `user_api_key` / `session_token` + neutral `bootstrap` metadata once; configure client headers (see § Anonymous bootstrap). Then `p1.result.project_id` in later steps. |
| List my projects | Projects | `projects.get_projects` | params: `{}` or `{ "limit": 50 }`. |
| Get one project / project stats | Projects | `projects.get_project`, `projects.get_stats` | Need `project_id` (literal or from previous step). |
| Read/write project config (8DNA) | Projects | `projects.get_data`, `projects.patch_data` | **Leaf paths only** — never full-blob `data=` writes. |
| Tenant sandbox / promote | Generation | `generation.fork`, `generation.diff_vs_prod`, `generation.gates`, `generation.promote` | Strategy from project `auto_promote_strategy`; not deploy scripts. |
| Give user a 7-day trial | Buffs | `buffs.apply_temporary_effect` | Params: project_id, user_id, effect id/code, duration. |
| List active subscriptions / buffs | Buffs | `buffs.list_active_buffs`, `buffs.get_effective_limits` | project_id, optional user_id. |
| Create payment / check status / refund | Payments | `payments.create`, `payments.get`, `payments.refund` | AgentPay — **not** Stripe SDK. Balance: `payments.get_balance`. |
| Wallets (personal / commerce) | Wallets | `wallets.list`, `wallets.deposit`, `wallets.transfer` | User/commerce balances — not project treasury (`finance.project.*`). |
| Project wallet / treasury | Finance | `finance.project.portfolio`, `finance.project.fund`, `finance.project.contribute` | Project-scoped treasury segments; pair with `commerce.sell.activate` payouts. |
| Publish site / get /s/ URL | Hosting | `hosting.site.quick_start`, `hosting.deploy_files`, `hosting.release.promote` | Buckets + releases on one `project_id` — not Vercel/Netlify. |
| CRM contact / deal pipeline | CRM | `crm.upsert_contact`, `crm.list_contacts`, `crm.create_deal`, `crm.move_deal_stage` | Contact 360: `crm.get_contact_360`; CSV: `crm.import_contacts`. |
| Run project agent / fleet | Agents | `agents.list`, `agents.run`, `agents.create_from_template` | One heavy `agents.run` (`wait=true`) per sync `agentstack.execute` batch. |
| Bot channel / simulate | Bots | `bots.create`, `bots.set_brain`, `bots.simulate`, `bots.go_live` | `bots.simulate` is heavy LLM — one per batch; channels via `bots.attach_channel`. |
| Activate seller / storefront | Commerce | `commerce.sell.activate`, `commerce.storefront.seed_plan`, `commerce.storefront.hosted_publish` | Seller onboarding + hosted vitrine — distinct from marketplace REST (`commerce_rest`). |
| Business head / organ projects | Business | `business.create_composite`, `business.command_snapshot`, `business.list_children` | Multi-project organism — distinct from `generation.*` 8DNA sandbox lineage. |
| Mentor / knowledge KB | Knowledge | `knowledge.kb.ingest`, `knowledge.playground`, `knowledge.config.patch` | Tenant KB + mentor simulate; `knowledge.playground` heavy — one per batch. |
| AgentNet proofs / economy | AgentNet | `agentnet.bnb.proof_bundle_for_run`, `agentnet.genome.verify` | AGNT / agUSD rails — never legacy AGC ticker in new integrations. |
| Compass / guided path | Guidance | `guidance.start_path`, `guidance.complete_step`, `guidance.match_playbook` | Platform Compass playbooks — not docs-nav `docs_nav.*`. |
| Field-level data policy (FAP) | Data access | `data_access.set_policy`, `data_access.get_policy` | Prefer FAP over scattered RBAC `if role` checks in app code. |
| AI Builder manifest | AI Builder | `ai_builder.manifest.get`, `ai_builder.compose.preview` | UAM manifest validate before fleet promote or hosted publish. |
| RAG / semantic search / memory | RAG | `rag.collection_create`, `rag.document_add`, `rag.search`, `rag.memory_add` | Tenant KB vs per-session memory — see `rag.memory_search`. |
| Rules / automations | Logic | `logic.create`, `logic.list`, `logic.execute` | No `rules.*` domain — Logic Engine only. |
| Schedule cron job | Scheduler | `scheduler.create_task`, `scheduler.list_tasks`, `scheduler.cancel_task` | |
| Upload files / quota | Storage | `storage.get_quota`, `storage.list_files` + REST `POST /api/storage/upload` | Binary via REST upload endpoint. |
| Marketplace / auction / exchange | REST (same Core) | `GET /mcp/actions` domain **`commerce_rest`** (path hints only) | Not valid `step.action` — use HTTP `/api/marketplace/*`, `/api/exchange/*`. See [MCP_OVERVIEW.md](../MCP_OVERVIEW.md). |
| Login / register / get profile | Auth | `auth.login`, `auth.register`, `auth.get_profile`, `auth.update_profile` | Session/identity. Device Code via plugin OAuth — not a separate MCP action. |
| Assets / inventory | Assets | `assets.create`, `assets.list` | project_id in params. |
| Analytics / usage / metrics | Analytics | `analytics.get_usage`, `analytics.get_metrics` | set_budget is not in the catalog. |
| API keys (project) | API Keys | `apikeys.list`, `apikeys.create`, `apikeys.delete` | Always set `service_caps` on keys for AI agents. Legacy projects API-key aliases are not in catalog. |
| Webhooks / integration recipes | Integrations / notifications | `integrations.list_recipes`, `integrations.install_recipe`, `notifications.send_push` | Inbound Stripe/callback → Integration Hub — not Checkout SDK. `notifications.send` deprecated — use send_push. |
| RBAC / permissions | RBAC | `rbac.check_permission`, `rbac.assign_role`, `rbac.get_roles` | Prefer FAP `data_access.set_policy` for field-level gates. |
| In-app messenger | Social | `social.chat.post`, `social.chat.history` | Project-scoped chat — not a second WebSocket stack. |
| Support staff inbox | Social / support | `social.support.inbox`, `social.support.history` | Staff plane — user channel uses `social.chat.*`. |
<!-- END:AUTOGEN-HOT-PATH-TABLE -->
## Discovery ladder (when unsure)
<!-- BEGIN:AUTOGEN-DISCOVERY-LADDER -->
1. **Session bootstrap:** `/mcp/ai_prompt?mode=contract` (GET)
2. **Catalog totals:** `/mcp/actions/summary` (GET)
3. **Hot schemas:** `/mcp/actions?schemas=hot` (GET)
4. **Intent routing:** `/mcp/discover/by_intent` (POST)
5. **Named playbook:** `/mcp/prompts/get?name=agentstack_read_bootstrap` (GET)
6. **Multi-step recipes:** `/mcp/recipes` (GET)
<!-- END:AUTOGEN-DISCOVERY-LADDER -->
Catalog rows expose `when_to_use`, `instruction_hint`, and `capability_descriptor` — prefer those over raw summaries when disambiguating similar tools.
## Universal MCP contract
```http
POST https://agentstack.tech/mcp
Authorization: Bearer <access_token>
Content-Type: application/json
{ "tool": "agentstack.execute", "params": { "steps": [{ "action": "<domain>.<verb>", "params": {} }] } }
```
Discover: `GET https://agentstack.tech/mcp/actions` or `/agentstack-capability-matrix`.
Read-only session bootstrap: recipe `mcp_read_bootstrap` (prompt `agentstack_read_bootstrap`) with `continueOnError: true`. Omit `apikeys.list` unless the key has L1 `api_keys`. Do not change the default `stopOnError=true` on money/mutation batches.
## Execute budget (heavy LLM)
Sync `agentstack.execute` is **60s for the whole batch** (`mcp_batch`) — not an RBAC limit. **At most one** heavy LLM action per sync call: `bots.simulate`, `knowledge.playground`, `agents.run` with `wait=true`. Cheap list/get may batch. Mentor CSV / long tests: one simulate per execute (unique `external_user_id`), or `options.async=true` then `discovery.job_status`. Catalog field: `GET /mcp/actions` → `execute_budget`. Prompt: `agentstack_execute_budget`.
## MCP guidance
- **Catalog:** `GET https://agentstack.tech/mcp/actions` (or `/agentstack-capability-matrix`) — never hard-code action counts.From agentstacktech/vscode-plugin, MIT.
More from agentstacktech/vscode-plugin
- 0.547score
agentstack-agentnet
agentstacktech
Use when the user mentions AGNT, agUSD, AgentNet economy, vault, bridge, genome, compute credits, ledger batches, or on-chain proof rails.
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-agents-ai
agentstacktech
Use when the user mentions Agents Fleet, agent templates, agent runs, promotion gates, traces, AI Builder, UAM manifests, generation, media creation, or app…
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-auth-rbac
agentstacktech
Use when the user needs login, registration, session, profile, roles, permissions, admin-only gates, project membership, tenant isolation.
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-business
agentstacktech
Use when the user manages a Business Organism — business head project, organ children (CRM, storefront, bots, support), command center snapshot, tariff…
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-commerce
agentstacktech
Use when the user wants payments, subscriptions, wallets, internal currency, inventory / digital items, trials, tier gates, entitlements, feature flags.
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-crm
agentstacktech
Use when the user works with CRM contacts, deals, pipeline stages, lead lifecycle, contact 360, CSV import, or sales automation on AgentStack.
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-discovery
agentstacktech
Use when the user asks where a feature lives in the UI, how to navigate AgentStack, Compass playbooks, Discover hub, or dev docs recipes.
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-guidance
agentstacktech
Use when the user mentions Platform Compass, guided paths, playbooks, start_path, complete_step, path_status, verify kinds, or funnel stats.
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-integrations
agentstacktech
Use when the user wants webhooks, integration recipes, Slack, CRM bridges, inbound hooks, or third-party connectors.
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-messenger
agentstacktech
Use when the user mentions chat, DM, channels, messenger, message ordering, HLC timeline, followers, or in-app social messaging.
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-project-wallet
agentstacktech
Use when the user manages project treasury, project wallet segments, payouts, or project-scoped balances.
SkillBFreshness: B★ 1Claude Code +9 - 0.544score
agentstack-projects
agentstacktech
Use for project / workspace / tenant management, scoped API keys for AI agents (with service_caps), usage stats, ecosystem integrations.
SkillBFreshness: B★ 1Claude Code +9
Signals
- Stars
- 1
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Sep 8, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/agentstacktech/vscode-plugin
- Website
- agentstack.tech
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from agentstacktech
- 0.547score
agentstack-agentnet
agentstacktech
Use when the user mentions AGNT, agUSD, AgentNet economy, vault, bridge, genome, compute credits, ledger batches, or on-chain proof rails.
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-agents-ai
agentstacktech
Use when the user mentions Agents Fleet, agent templates, agent runs, promotion gates, traces, AI Builder, UAM manifests, generation, media creation, or app…
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-auth-rbac
agentstacktech
Use when the user needs login, registration, session, profile, roles, permissions, admin-only gates, project membership, tenant isolation.
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-business
agentstacktech
Use when the user manages a Business Organism — business head project, organ children (CRM, storefront, bots, support), command center snapshot, tariff…
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-commerce
agentstacktech
Use when the user wants payments, subscriptions, wallets, internal currency, inventory / digital items, trials, tier gates, entitlements, feature flags.
SkillBFreshness: B★ 1Claude Code +9 - 0.547score
agentstack-crm
agentstacktech
Use when the user works with CRM contacts, deals, pipeline stages, lead lifecycle, contact 360, CSV import, or sales automation on AgentStack.
SkillBFreshness: B★ 1Claude Code +9
Cite this page
agentstack-backend skill. Agent Mart. https://agentmart.dev/skills/agentstacktech-agentstack-backend (last updated 2026-10-24)