prismeai-mcp

Index of the Prisme.ai plugin skills plus essential Prisme.ai environment context (environments, workspaces, tools, event schema). Use this when starting work on anything Prisme.ai-related and unsure which skill applies, or to discover what is available.

Edit a React app embedded in a Prisme.ai workspace (starter-spa pattern). Bootstraps from github.com/prismeai/starter-spa if the workspace has no app yet, then modifies src/ (React) and/or automations/ (DSUL), builds the CJS bundle if needed, and pushes to the workspace via the prisme-ai-builder MCP (automations) + direct fetch (bundle/files/config). Use when the user says "edit the X app", "modifier l'app de <workspace>", "/workspace-page-implement <workspace> <change>", or when src/App.tsx + scripts/deploy.mjs are detected in a workspace folder.

Build, edit, test and deploy a Prisme.ai Agent Factory agent whose definition is git-versioned inside a workspace. The prompt lives in AGENT.md; model/profile/capabilities/app+mcp/tests live in agent.yml. Use when the user wants to scaffold a new agent, change an agent's prompt or capabilities, list/search the platform's available capabilities, install and wire an App+MCP connector as an agent tool, manage the test battery (list/add/edit tests, run the suite), choose or change the deployment organization, or deploy the agent (create/update + publish + push the workspace) to sandbox or prod. Triggers include "crée un agent", "modifie le prompt de l'agent", "ajoute une capacité", "quelles capacités sont dispo", "installe l'app+mcp X", "ajoute/lance les tests", "déploie dans l'organisation X", "change d'organisation", "déploie l'agent", "/agent-builder ...".

Add A2UI (Agent-to-UI) surfaces to a Prisme.ai MCP workspace so that LLM agents can render interactive UI through MCP tool calls. Scaffolds an MCP endpoint (or extends an existing one) with tools whose outputs include a __surface payload (components, data_model, actions) that the host UI renders using the prisme://blocks/v1 catalog. Use when the user says "ajoute des surfaces A2UI", "expose une UI via MCP", "/agent-implement-a2ui <workspace> <surfaces>", or wants the agent to draw cards/forms/tables/buttons inside a chat.

Create a Prisme.ai bootstrap workspace that provisions an Agent Factory agent, attaches Storage-backed file or URL sources, seeds Agent Evaluations cases, runs evaluation, and validates the setup through the supported product apps Agents, Storage, and Evaluation.

Build or audit Prisme.ai App+MCP consumer DSUL workspaces that consolidate tests for a target App surface and its MCP server through Agent Factory agents. Use when implementing, recreating, or assessing connector consumer workspaces.

Generate public documentation (MDX) for a Prisme.ai App+MCP connector, mirroring the structure of existing pages (gryzzly.mdx, data-galaxy.mdx, …) in the prismeai/docs repo. Produces one `apps-store/marketplace/connectors/<slug>.mdx` page and inserts a matching `<Card>` into `overview.mdx`. Use when the user says "document the X app+mcp", "écris la doc de X", "/app-mcp-document X", or similar. Assumes the source workspace already exists locally (scaffolded by `/app-mcp-implement`).

Find every Prisme.ai App+MCP connector workspace (by the `app-mcp` label) and check each one against the current `/app-mcp-implement` skill templates + best-practice rules, reporting drift and applying approved fixes. The connectors are the workspaces scaffolded by `/app-mcp-implement`; this skill keeps the whole fleet in sync when a template or trap-fix lands. Use when the user says "update all app-mcp connectors", "vérifie que les connecteurs sont à jour", "propage le fix de template à tous les connecteurs", "/app-mcp-fleet-sync", or similar.

Build a brand-new Prisme.ai App+MCP connector for a third-party SaaS using the tenant-context model (no HMAC), an entity-grouped registry driven by a generated OpenAPI spec, multi-mode auth (API key / OAuth2 client-credentials / OAuth2 per-user PKCE / JWT service-account) resolved by buildAppAuth, a central platform OAuth client (token-service model — zero-config `oauthCentral` tenant mode + maintainer view in the config SPA), per-user OAuth tokens resolvable from cron, and a model-B config SPA. Reproduces the validated `salesforce-next` build (+ `google-workspaces` central OAuth) for any service. Use when the user says "build an app+mcp for X", "créer une app+mcp pour X", "implémente un connecteur X". Everything needed is here + in `reference/`.

End-to-end smoke-test the tools exposed by a Prisme.ai App+MCP workspace, then hand off to app-mcp-build-consumer to consolidate the successful coverage into a consumer test workspace. Asks for the workspace and environment, collects the credentials required by the app's config schema, lists every MCP tool, then executes them one by one, diagnosing and fixing errors in the workspace until the full suite passes. Supports static tokens, Basic, OAuth2 client-credentials, AND OAuth2 authorization-code (PKCE) when the user supplies a pre-configured tenant workspace whose user already completed the browser auth — in that mode we call `ensureAuthentication` first to mint a `$secret:` accessToken ref and inject it into `routeToolCall`/`tool-*` directly. Use when the user says "test this app+mcp", "tester les tools du MCP X", "/app-mcp-test", or similar.

Tracer et diagnostiquer une exécution Prisme.ai via le flux d'events (activity feed). À partir d'un correlationId, d'un slug d'automation, d'un message d'erreur ou d'un symptôme ("l'agent boucle", "le tool call ne part jamais", "500 intermittent"), reconstruit la chaîne d'exécution avec `search_events`, localise l'automation+étape fautive et propose la cause racine. Read-only. Déclencheurs : "debug cette erreur", "trace ce correlationId", "pourquoi cette exécution a échoué", "activity feed", "search events", "/debug-events".

Référence des pièges DSUL & Custom Code Prisme.ai qui passent `validate_automation` mais explosent au runtime (InvalidExpressionSyntax, Function not found, comparaisons silencieusement fausses, OOM…). À consulter AVANT d'écrire ou de débugger toute automation, expression `{% %}`/`{{ }}`, fonction Custom Code, condition, ou requête Collection. Déclencheurs : "pourquoi mon automation 500", "InvalidExpressionSyntax", "Function not found", "ma condition ne matche jamais", "ternaire DSUL", "/dsul-rules".

Investigate Prisme.ai automations, apps, events, and documentation. Use when debugging executions (correlationId, activity feed, logs), inspecting automations, or searching Prisme documentation. Requires the workspace name or ID, the environment (sandbox or prod), and the details to investigate.

Verify that an implementation fully satisfies a ticket/spec. Use AFTER implementing a feature, passing the ticket path or content. Run BEFORE code review. Produces a requirements checklist with IMPLEMENTED/PARTIAL/MISSING statuses.