odoo-coder
sonnetShip production Odoo Python on the first pass — every field verified against the index, every ORM chain validated, and a failing test written before the implementation.
validate_depends · resolve_orm_chain
Odoo AI Agent Team is a Claude Code plugin: 41 pre-built skills, 7 specialist agents, and 9 slash commands that run the whole delivery pipeline — intake, design, code, review, ship — with every step verified against an indexed knowledge graph of 184,000+ fields across Odoo 8 to the latest version. One install. No hallucinated field names.
See the 25 MCP tools and 7 resources it runs, or the 40-task accuracy benchmark.
41
Skills
7
Specialist agents
9
Workflow commands
9
Personas covered
When an AI coding assistant guesses without grounding, each task accumulates errors. An ungrounded coder
suggests account.invoice — deprecated
since v13. An ungrounded reviewer approves it. An ungrounded tester misses it. One wrong field name compounds
through design, code, review, and ship.
account.invoice.search(...) — model removed in v13@api.depends('amout_total') — typo, silent failsale.orderamount_total with no idea what breaks downstreamaccount.move.search(...) — resolved against the index@api.depends('amount_total') — path validatedAnswer accuracy on 40 real Odoo tasks
43% 95%
ungrounded vs grounded, same model on the same task set
Tested 2026-05, Claude claude-sonnet-4-5. See the 40-task benchmark methodology.
Odoo AI Agent Team provides seven specialist agents: an architect (Opus-tier, design-gate), five engineering agents (Sonnet-tier), and two UI agents with live browser tools. Each agent queries the indexed knowledge graph on every run — not its training memory. They cross-link through 25 MCP tools.
Get an AI architect that designs the inheritance axis, override chain, and impact matrix before your developers touch a single file — grounded in the indexed codebase, not guesswork.
tools: model_inspect · find_override_point · impact_analysis
Ship production Odoo Python on the first pass — every field verified against the index, every ORM chain validated, and a failing test written before the implementation.
validate_depends · resolve_orm_chain
OWL components, QWeb templates, and SCSS overrides that render on-theme on the target Odoo version — import paths and design tokens grounded in indexed source, not training memory.
resolve_stylesheet · api_version_diff
Every finding is evidence-backed with indexed output and line citation — not a generic linter, but an Odoo-aware review that catches the bugs that slip past conventional static analysis.
lint_check · impact_analysis
Proven root cause, not a plausible guess — the debugger cannot fill its Output Contract until it has actually toggled the cause on and off.
resolve_orm_chain · find_override_point
Diagnose a blank OWL screen or invisible CSS in minutes — the agent reads both live computed styles and the indexed stylesheet chain to pinpoint the exact file and selector, not just "clear cache and try again".
browser tools · resolve_stylesheet
A six-lens UI verdict with Lighthouse scores, accessibility findings, and design-token reality check — grounded in both the live rendered screen and the indexed stylesheet source, not a visual eyeball.
browser tools · find_style_override
Odoo AI Agent Team's 41 skills cover the full Odoo development lifecycle — from architecture and coding through review, testing, deployment, sales, and marketing. Describe an outcome in plain English ("discount can never exceed 20% of unit price") and the right specialist auto-routes to the right agent and the right tool.
Every skill grounds its output against the index — field names, editions, API signatures, and override chains. Skills can be assembled into 12 declarative multi-phase workflows behind the slash commands below.
Odoo AI Agent Team ships 9 slash commands for when you want a named multi-phase pipeline instead of a single skill. Each command chains skills and agents in a fixed sequence — from a full bid response to a git-wave that lands multiple changes safely.
/odoo-respond-bid Generate a complete bid response package from raw prospect input through six gated phases: discovery synthesis, gap analysis, capability proof, objection pre-emption, and proposal draft.
/odoo-draft-followup Draft a follow-up email for a stalled or at-risk deal, with explicit save-to-disk confirmation before any file is written — never auto-sends.
/odoo-summarize-discovery Synthesize raw meeting or discovery notes into a structured customer profile covering business context, pain points, goals, and product fit.
/odoo-respond-bid — generate a complete bid response package from raw prospect input through six gated phases
/odoo-draft-followup — draft a follow-up email for a stalled or at-risk deal, with explicit save-to-disk confirmation before any file is written — never auto-sends
/odoo-summarize-discovery — synthesize raw meeting or discovery notes into a structured customer profile covering business context, pain points, goals, and product fit
/odoo-position-feature — generate positioning copy for a specific odoo feature across four phases
/odoo-plan-upgrade — generate a comprehensive odoo upgrade plan from source version to target version across four phases
/odoo-setup — one-shot idempotent setup
/odoo-run-brl — process a business requirement list (brl) of any size into a four-way classified, costed, dependency-ordered plan with rtm export (rtm
/odoo-produce-video — produce a multi-scene odoo demo video across three gated phases
/odoo-run-wave — kick off depth-0 multi-subagent git-wave orchestration
Every agent in Odoo AI Agent Team makes its claims verifiable. When the coder suggests
amount_total, it verified that field
exists on sale.order at your target
version first. When the reviewer flags a dependency, it ran
validate_depends against the indexed
graph. Graph lookup does not hallucinate: either the field exists in the index at that version, or it does not.
You describe an outcome in plain language
The right skill auto-routes to a specialist agent
The agent queries the indexed knowledge graph
A verified answer — cited to real source, not memory
One install pulls the companion connector automatically. Agents read 25 MCP tools and 7 resources over a single namespace, against an index of 12,400+ models and 184,000+ fields — cross-repo, cross-version, inheritance already resolved, covering Odoo 8 to the latest version.
>_ validate_depends(sale.order, ['amout_total'])
× field not found — did you mean amount_total?
>_ validate_depends(sale.order, ['amount_total'])
✓ valid path verified against the index
Odoo AI Agent Team is not a replacement for your IDE language server or local context tools. It is the orchestration layer that fires the right indexed query at the right moment. Here is how it relates to the tools developers already use — see the 25 MCP tools and 7 resources it runs.
| Tool | What it is | Relationship |
|---|---|---|
| Odoo AI Agent Team | Claude Code plugin: 41 skills, 7 agents grounded on the index | This plugin |
| Odoo Semantic MCP | The knowledge graph — 25 tools, 7 resources | Auto-installed as a dependency |
| odoo-ls | IDE language server, single local checkout | Complementary — checks syntax in your editor |
| Akaidoo | Local context loader, single checkout | Complementary — loads local context |
| DB bridge MCPs | Live Odoo record reader/writer | Different use case — runtime data |
Odoo AI Agent Team covers 9 roles across the full Odoo lifecycle. The developer family is the primary audience; business personas get outcome-led skills that ground every claim against the indexed source. Browse the examples showcase to see each role in action.
Writes idiomatic Odoo backend (Python/XML) or frontend (JS/OWL) code and debugs runtime failures using odoo-coding, odoo-debug, odoo-test-writer, odoo-security-audit, and odoo-perf-audit.
Finds correct override points, audits deprecated APIs before an upgrade, and validates deployments using odoo-override-finding, odoo-deprecation-audit, and odoo-deploy-checklist.
Reviews PRs or audits patches for ORM misuse, inheritance anti-patterns, security holes, and N+1 queries using odoo-code-review with specialist agents odoo-code-reviewer and odoo-backend-debugger.
Reviews live Odoo screens across six lenses, debugs broken renders, catches visual regressions, records demos, and runs the full QA pipeline using odoo-ui-review, odoo-visual-regression, odoo-demo-recording, and odoo-qa-suite.
Verifies feature availability, builds gap matrices, produces evidence for proposals, and classifies and costs BRL requirements at scale using odoo-feature-check, odoo-gap-analysis, odoo-capability-proof, odoo-addon-diff, and odoo-brl.
Produces ACA-structured objection responses, risk-scored follow-up emails for stalled deals, prospect profile syntheses, and RFP compliance matrices using odoo-objection-handling, odoo-deal-followup, odoo-discovery-summary, and odoo-rfp-response.
Creates content around Odoo features — blog posts, slide decks, social copy, and multi-channel campaign plans — in marketing-ready language using odoo-feature-highlights, odoo-content-draft, odoo-campaign-plan, and odoo-competitive-brief.
Produces executive risk overviews of customizations, structured customization inventories, competitor capability snapshots, and customer health scores using odoo-risk-overview, odoo-customization-inventory, odoo-competitive-brief, and odoo-customer-health.
Universal front door for every persona — bootstraps project context, routes any plain-language intent to the right specialist or workflow using odoo-intake, odoo-onboarding, run-driver, and workflow-chaining.
Install Odoo AI Agent Team in three commands. The first registers the Viindoo marketplace, the second installs the plugin (auto-pulling its companion connector), and the third connects to the knowledge graph with your API key. The full setup guide covers every AI tool.
Register the Viindoo plugin marketplace.
Auto-pulls the MCP connector dependency.
Paste your URL and API key from the dashboard.
$ claude plugin marketplace add Viindoo/claude-plugins --scope user
$ claude plugin install odoo-ai-agents@viindoo-plugins --scope user
# inside Claude Code:
/odoo-semantic-mcp:connect # paste URL + API key
On 40 real-world Odoo coding tasks tested in 2026-05 with Claude claude-sonnet-4-5, AI grounded with OSM answered correctly 95% of the time. The same model without OSM answered correctly 43% of the time. The gap is largest on inheritance chain traversal and field version questions.
Odoo AI Agent Team is a Claude Code plugin that provides 41 pre-built AI skills, 7 autonomous specialist agents, and 9 slash commands. Every agent is grounded on the Odoo Semantic MCP knowledge graph, which indexes 12,400+ models and 184,000+ fields across Odoo 8 to the latest version. Unlike ungrounded AI, agents verify field names and inheritance chains against the indexed source before suggesting code.
The plugin supports Odoo 8 through the latest version. Recent major versions are actively maintained with regular index updates. Versions v8 through v13 are available as legacy with no active index updates. The newest in-development release is indexed from its development branch and is not recommended for production use.
Run two commands in your terminal: (1) claude plugin marketplace add Viindoo/claude-plugins --scope user (2) claude plugin install odoo-ai-agents@viindoo-plugins --scope user. Then inside Claude Code run /odoo-semantic-mcp:connect and enter your API key from odoo-semantic.viindoo.com. Total setup time: under 2 minutes.
Yes. The free tier provides 30 MCP queries per day with no credit card required. Paid plans with higher daily limits are available - see the pricing page for current options. All plans include the full set of 41 skills, 7 agents, and 9 commands; the free tier is limited only by daily query count. See all plans.
odoo-ls is an IDE language server for syntax checking and autocompletion in a single local Odoo checkout. Odoo AI Agent Team is a Claude Code plugin that runs 7 specialist AI agents grounded on a cross-version knowledge graph covering 12 Odoo versions. Both can be used together: odoo-ls checks syntax in your editor while Odoo AI Agent Team handles semantic questions — what breaks if I change this field, what is the override chain, is this ORM path valid across versions.
Odoo AI Agent Team is optimized for Claude Code. The underlying OSM knowledge graph (Odoo Semantic MCP) works with any MCP-compatible tool including Cursor, Codex CLI, Gemini CLI, VS Code 1.99+, Windsurf, JetBrains AI Assistant, and Continue.dev. The 41 skills and 7 agents are Claude Code-specific; other tools access the 25 MCP tools directly.
In a benchmark of 40 real-world Odoo coding tasks tested in 2026-05 with Claude claude-sonnet-4-5, AI grounded with OSM answered correctly 95% of the time. Without OSM, the same model answered correctly 43% of the time. The 52-point gap is largest on inheritance chain traversal and field version questions, where ungrounded AI hallucination rates are highest.
The plugin
Free Claude Code plugin. 41 skills, 7 agents, 9 commands. Installs with one command — nothing to pay for the plugin itself.
claude plugin install odoo-ai-agents@viindoo-plugins
The service it connects to
The indexed graph the plugin queries. Free tier: 30 queries/day, no credit card. Paid plans for professional use.
See full pricingInstall the agent team in under two minutes and ground every Odoo answer in indexed source.