By jjackson
AI Connect Engine — orchestrates the ACE lifecycle from idea through app building, Connect setup, LLO management, and closeout
Generate a new ace-web video program (spec.yaml) from a Connect program page URL using the 60s-campaign-overview template
Render an ace-web connect-videos program to MP4 locally (bare-metal Mac, fast) — from a local spec + master clip, or an existing Drive program
Mint a per-human personal access token (PAT) for ace-web via a gh-style loopback browser flow, write it to the local .env, and verify it round-trips. Replaces the deployment-wide ACE_E2E_AUTH_TOKEN shared secret. One-time per machine; re-run to rotate.
Add a Dimagi teammate to a Connect workspace (organization) by email so they can see what ACE runs there. Defaults to adding you (your git email) to ai-demo-space. @dimagi.com only.
Manual fallback login for Connect (connect.dimagi.com). Opens a headed Playwright browser so the user can sign in via OAuth-with-CommCareHQ (covers MFA / SSO edge cases), then saves the resulting session state to ~/.ace/connect-session.json. The Playwright backend auto-logs-in via ACE_HQ_USERNAME/ACE_HQ_PASSWORD by default — only use this command when those credentials are unavailable or the account requires interactive auth.
Orchestrates opportunity closeout: invoice processing, LLO feedback collection, learnings summary, and overall cycle grading. Triggered when the opportunity reaches its end date.
Phase 3 of the ACE lifecycle: translate the approved PDD into Learn and Deliver apps via Nova, deploy them to CommCare HQ, and test.
Orchestrates Connect platform setup for an ACE opportunity: program creation, opportunity shell, verification flags, and payment units. Now atom-driven via the ace-connect MCP (no HITL).
Phase 9 of the ACE lifecycle: execute the awarded LLO's run of the opportunity — onboarding, UAT, go-live, and recurring monitoring. Phase 9 entry is gated on `phases.solicitation-management.products.selected_llo.org_slug` being populated in the current run's run_state.yaml by Phase 8's solicitation-review skill.
Phase 1 of the ACE lifecycle: iterate an initial idea into an approved Program Design Document (PDD). The PDD is the formal design artifact every downstream phase builds on. Phase 2 (scenarios-and-acceptance) derives test prompts and app journeys from the approved PDD.
Add a Dimagi teammate to a Connect workspace (organization) by email, so they can see the programs/opportunities ACE runs there. Thin wrapper over the `connect_add_org_member` atom: enforces a @dimagi.com guard, defaults the workspace to `ai-demo-space` and the email to the session's own git identity ("add me"), invites via Connect's membership form, and verifies by member-table read-back. Anyone running ACE can invoke it; ACE performs the add as its own org-admin identity.
Verify every form in a Nova-built Learn or Deliver app has the right CommCare Connect markers, auto-fix via Nova edits, loop until clean.
Upload Nova-built Learn + Deliver apps to CommCare HQ as draft builds via /nova:upload_to_hq. Captures HQ app IDs and writes a deploy summary.
Attach display-only images to Connect app questions where they meaningfully help FLWs. Manual gate; not part of /ace:run.
Verify every Learn + Deliver build was actually released so Connect can read deliver units. Provisional rubric pending 3+ real releases.
Admin access level
Server config contains admin-level keywords
External network access
Connects to servers outside your machine
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Uses power tools
Uses Bash, Write, or Edit tools
Uses power tools
Uses Bash, Write, or Edit tools
Runs pre-commands
Contains inline bash commands via ! syntax
Runs pre-commands
Contains inline bash commands via ! syntax
Bash prerequisite issue
Uses bash pre-commands but Bash not in allowed tools
Bash prerequisite issue
Uses bash pre-commands but Bash not in allowed tools
Orchestrates the ACE lifecycle for Connect opportunities, from idea through app building, deployment, LLO management, and closeout.
ACE is a Claude Code plugin with the same architecture as canopy: agents orchestrate skills, skills are prompt-based capability definitions, and MCP servers provide programmatic access to external systems.
Before /ace:setup will succeed, you need:
npm (https://nodejs.org). Apple Silicon: brew install node.op). Mac: brew install 1password-cli. Other platforms: https://developer.1password.com/docs/cli/get-started/.
op signin --account dimagi.1password.com interactively when /ace:setup prompts.dimagi.1password.com, with read access to the AI-Agents vault. Ask Jon (or whoever owns the vault) to grant access — without it, op inject will fail with cryptic permission errors. The full set of items ACE reads from AI-Agents: ACE - Google Service Account (Document), ACE - Open Chat Studio, ACE - CommCareHQ, ACE - Connect Labs, Content Generator API, connect-test-user, plus two UUID-referenced API-key items (linked from .env.tpl).jjackson/ace (currently public; no extra step needed). gh auth login is not required unless you plan to push.openchatstudio.com, connect.dimagi.com, commcarehq.org, labs.connect.dimagi.com, googleapis.com. Corp VPNs and proxies sometimes block these — /ace:doctor reports an explicit status per host.npx playwright install chromium.Mobile (Phase 6 only) is currently Mac-only. Phases 1–4 and 6–9 work on Mac, Linux, and Windows. Phase 6 (mobile screenshot capture against an Android emulator) has only been live-validated on macOS Apple Silicon — the Linux/Windows installer commands in commands/mobile-bootstrap.md exist but haven't been tested end-to-end. If you're on Windows, you can run everything except Phase 6; ask Jon for the workaround.
/plugin marketplace add jjackson/ace # Add the ACE marketplace
/plugin install ace@ace # Install the plugin
/ace:setup # Install deps + verify the service-account key
/ace:doctor # Sanity-check everything
/ace:run # Run full lifecycle (smart defaults)
/ace:run <opp-name> --mode review # Run full lifecycle for a named slug
/ace:run --dry-run # Test without side effects
/ace:run --sandbox # Route to staging endpoints
/ace:step <skill-name> <opp-name> # Run a single step
/ace:status # Show all opportunities
/ace:eval <opp-name> --mode deep # Umbrella eval (aggregates verdicts/*)
/ace:docs # Generate playbook
/ace:update # Pull the latest release from GitHub
The commands above are only the happy path once everything is configured. For a fresh install (or a new machine), run through this checklist top-to-bottom. Stop at any step that fails — the next step won't work.
Install the plugin — /plugin marketplace add jjackson/ace then
/plugin install ace@ace.
Run /ace:setup — does almost everything for you in one shot:
verifies Node + op + 1Password signin, fetches the GWS service-account
key from the AI-Agents vault into ${CLAUDE_PLUGIN_DATA}/gws-sa-key.json,
runs npm install, injects .env from 1Password, and finishes by
running /ace:doctor.
op isn't signed in, the script prints the exact
! op signin --account dimagi.1password.com line — type it in the
chat (the ! prefix runs the command in this session), then re-run
/ace:setup.AI-Agents so you can pick the right one with
ACE_GWS_KEY_OP_DOC=.... Worst case, ask Jon for the JSON and drop
it manually at the path the script printed.Authenticate to OCS — /ace:ocs-login opens a headed browser so you
can sign in (SSO/MFA included). Session state is saved to
~/.ace/ocs-session-<team>.json for headless reuse.
OCS_USERNAME /
OCS_PASSWORD resolved into .env (they do by default for
[email protected]), the MCP backend auto-logs-in on first call.
/ace:ocs-login is the manual fallback for SSO/MFA edge cases.npx claudepluginhub jjackson/ace --plugin aceGenesis (Forerunner) - AI-Powered Executive Director Automation Platform. Analyze organizational responsibilities, score automation potential with 6-factor algorithm, generate LangGraph workflows, and deploy 11 specialized agents for trade associations and nonprofits.
Security and privacy expertise with MCP security, OAuth, and GDPR/CCPA compliance
Expert skills for designing, building, and deploying Make.com automation scenarios — for Claude Code and other AI agents
Create, customize, and manage plugins tailored to your organization's tools and workflows. Configure MCP servers, adjust plugin behavior, and adapt templates to match how your team works.
Transform Claude Code into a structured development platform with 29 /sc: commands, 23 specialized agents, 7 behavioral modes, and MCP server integration
Advanced multi-agent coordination platform with task orchestration, performance monitoring, and workflow optimization. Features hooks for agent lifecycle events and MCP server for state management.