Instructions file
Instructions for Viindoo/odoo-mcp-client, covering claude.md, what this repo is, commands, architecture you must understand before editing and ssot generator - never hand-edit generated regions.
Plugin Claude Code
Safe, scale-aware git + GitHub toolkit for AI agents - one front-door skill that runs git/github work in a delegated context so the caller stays clean and code is never lost. 1 skill + 4 agents (read-only surveyor, write operator, GitHub operator, pipeline lead). Backup-before/verify-after on destructive ops…
Agent
Use this agent when an orchestrator needs to EXECUTE a local git mutation safely - integration (fetch, pull-rebase, merge, cherry-pick, forward-port, backport, branch/tag/worktree, non-force push) OR a destructive history rewrite (interactive rebase, squash, split, amend, reset, filter-repo, force-with-lease push). It…
Agent
Use this agent when a git/github change is COMPLEX or LARGE - more than 500 files, more than 10,000 lines, a multi-commit history rewrite, or a thousand-file backport/forward-port - and the whole phased pipeline should run BELOW the caller so the caller's context stays pristine. It is the ONLY agent that spawns…
Agent
Use this agent when an orchestrator needs READ-ONLY git cognition - mapping, analyzing, evaluating, or proving - over a repo, a diff range, or a commit history, without mutating anything. Typical triggers include a phased-pipeline P1 map pass (cluster thousands of changed files by module), a P2 per-cluster evaluation…
Agent
Use this agent when an orchestrator needs a GitHub API operation - the PR lifecycle (create, read, review, merge), issue triage (read, write, comment), branch/commit/tag queries, file reads, code search, releases, CI status, or fork -> PR upstream. It uses the GitHub MCP tools as PRIMARY and the gh CLI as fallback…
Skill Claude CodeCodex
Universal front door for ALL git and GitHub work: it ROUTES the request and DISPATCHES a specialist agent to execute it in a delegated context; it never runs the churn itself, so the caller stays clean and code is never lost. Trigger whenever a request touches git or GitHub in ANY way, even casual phrasing…
Plugin Claude Code
Odoo AI workforce toolkit - 51 skills + 26 specialist agents + 8 workflow commands across engineering, sales, marketing, strategy, setup, visual UI testing, and automated multi-module git integration. Pairs with the odoo-semantic-mcp server for indexed-codebase grounding (auto-installed). Includes browser-automation…
MCP server Claude CodeCodexCursor +2
Lets the agent inspect and automate a live Chrome instance through the DevTools protocol: performance traces, network, console. Runs locally from the chrome-devtools-mcp npm package.
Agent
Use this agent when the main agent needs to write production-ready Python/XML Odoo backend code - computed fields, ORM overrides, constraints, migration scripts. It implements to a RED test the odoo-test-writer teammate already authored (it does NOT author tests). Dispatched by the odoo-coder per-node coordinator as…
Agent
Use this agent when the main agent needs to diagnose a Python/ORM/server-side Odoo runtime failure to a PROVEN root cause - data-state bugs, Expected singleton, AccessError (ir.model.access vs ir.rule), compute/onchange/constraint/ORM-hook-order bugs, tracebacks, module-load/migration/ParseError. OSM-only (no…
Agent
Use this agent when main agent needs to review existing Odoo Python/JS/XML/OWL code for bugs, convention violations, security issues, N+1 queries. Produces CRITICAL/HIGH/MED/LOW findings + corrected version.
Agent
Use this agent as the COORDINATOR the odoo-coding skill launches for EVERY work node (backend-only, frontend-only, or full-stack; a node may name one module, part of one, or several). For its node it computes an INTERNAL work-item (WI) breakdown - splitting the node's changes into 1..N WIs by DISJOINT file sets …
Agent
Use this agent when an orchestrator needs a structured business behavior (nghiệp vụ) / intent (ý đồ) / expected-outcome / acceptance-criteria comparison between two code states - read-only, writes only its findings file. Typical triggers include odoo-git-rebase P3 dispatching a cluster-level comparison of the feature…
Agent
Use this agent when a documentation package spanning MORE THAN ONE Odoo module needs a dependency-aware execution plan before any instance is provisioned - it partitions the in-scope modules into dependency clusters, allocates instances by DAG SHAPE (a linear chain shares one instance and installs incrementally…
Agent
Use this agent when the odoo-doc-walkthrough skill (or another caller) needs happy-path usage scenarios authored for an Odoo module - structured walkthroughs (role, precondition, step list, expected outcome) grounded in the module's ACTUAL behavior via OSM, ready to drive the odoo-doc-illustration skill CAPTURE MODE…
Agent
Use this agent when the doc pipeline needs to resolve the documentation target, map it to Odoo modules, and produce a compact scope block before dispatching the odoo-doc-illustration skill. Typical triggers include the odoo-doc-illustration skill receiving a TARGET=repo: instruction for a multi-module scan, a…
Agent
Use this agent when the odoo-doc-feature-map skill (or another caller) needs a complete, machine-readable capability inventory for ONE Odoo module done in its OWN context so the orchestrator stays context-clean. It enumerates every user-visible feature a module ships - menus, views, models, actions, key fields, roles…
Agent
Use this agent when main agent needs to write production-ready Odoo frontend code (JavaScript, OWL, QWeb, SCSS) for any supported version - legacy web.Widget/AbstractField/odoo.define() (v8-v14) or OWL 2.x patch()/useState/useService (v15+). Produces complete files + manifest wiring. It implements to a RED JS test the…
Agent
Use this agent when the odoo-gap-analysis skill (or another caller) needs the heavy gap-analysis work for ONE requirement cluster done in its OWN context, so the orchestrator/main stays context-clean. It classifies each requirement against Odoo standard functionality - coverage (full/partial/none), classification…
Agent
Use this agent when the main agent needs to design and generate an Odoo module icon asset - composing a brand-aware, version-correct SVG then rasterizing it to static/description/icon.png (256x256 PNG). Typical triggers: "design an icon for this module", "generate icon.png for ", "create the app icon for this addon"…
Agent
Use this agent when the forward-port pipeline needs to resolve a category-3 ambiguity - the target clean-tip module descriptor shows installable: True for a module, but that manifest was not touched by any commit in the forwarded commit range, so the SOURCE-side manifest history must also be read to confirm whether…
Agent
Use this agent when a human OR another agent needs a live Odoo instance built, dropped, or driven for ANY series from v8 onward - create or drop a database through Odoo, init or update modules, run tests, ensure an instance is up, or report status - and wants structured metadata back including a persistent log path.…
Agent
Use this agent when the main agent needs to extract the business intent, purpose, and behavioral contract from one or more Odoo commits belonging to the SAME module - separating what behavior each commit was designed to produce from its implementation details. Read-only, one MODULE per instance: odoo-forward-port's P1…