gisce-erp

name	gisce-erp
description	Interact with the GISCE ERP service (based on OpenERP) to query, create, update and delete records in any model. Use this skill whenever the user asks to read data from the ERP, modify ERP records, explore the ERP schema, run business methods, or integrate with the GISCE/OpenERP backend. Trigger keywords: ERP, GISCE, OpenERP, model, partner, invoice, factura, contract, pòlissa, comptador, meter, supply point, CUPS, search, read, write, create, unlink.
allowed-tools	shell bash

GISCE ERP Client Skill

Use the pygisceclient CLI command to interact with the GISCE ERP. Every call follows this pattern:

pygisceclient \
  --url <PROTOCOL_URL> \
  --database <DB> \
  --user <USER> --password <PASS> \
  (--model <MODEL> | --service <SERVICE>) \
  --method <METHOD> \
  [--service-auth auto|raw|authenticated] \
  --args '<JSON_ARRAY>' \
  --kwargs '<JSON_OBJECT>'

Output is always JSON on stdout. Errors go to stderr with a non-zero exit code.

---

1. Connection & Authentication

Protocols

The --url must include a protocol prefix:

Prefix	Transport
https+restapi://	REST API (Flask)
https+msgpack://	MsgPack API
https+xmlrpc://	XML-RPC
https+xmlrpc-wst://	XML-RPC with server transactions

Use http+… variants for non-TLS connections. Add --no-verify to skip SSL certificate validation.

Auth modes (mutually exclusive)

• User/password: --user admin --password secret
• Token: --token <API_TOKEN>

--password is required when --user is provided.

---

2. ERP Architecture Overview

The GISCE ERP uses an ORM layer over PostgreSQL. Every business entity is a model identified by a dotted name (e.g. res.partner, account.invoice). Models expose standard CRUD methods plus custom business logic. All protocols (REST, MsgPack, XML-RPC) dispatch through the same ORM pool, so the same models and methods are available regardless of protocol.

---

3. Standard ORM Methods

fields_get — Discover the schema of a model

Always call fields_get first on any model you haven't seen before to learn which fields exist, their types, and constraints.

pygisceclient ... --model res.partner --method fields_get

Returns a dict mapping field names → definitions (type, string, relation, required, readonly, …).

Shortcut to list just the field names:

pygisceclient ... --model res.partner --method fields_get | jq 'keys'

search — Find record IDs

pygisceclient ... --model res.partner --method search \
  --args '[[["active", "=", true], ["customer", "=", true]]]'

Positional args in order: domain, offset (int, default 0), limit (int), order (string), context (dict), count (bool).

Returns a JSON array of integer IDs (or a count if count=true).

search_count — Count matching records

pygisceclient ... --model res.partner --method search_count \
  --args '[[["active", "=", true]]]'

read — Fetch record data

pygisceclient ... --model res.partner --method read \
  --args '[[1, 2, 3], ["name", "vat", "city"]]'

Positional args: ids (list[int]), fields (list[str] or null for all).

Returns a list of dicts.

write — Update records

pygisceclient ... --model res.partner --method write \
  --args '[[42], {"name": "New Name", "city": "Barcelona"}]'

Positional args: ids (list[int]), vals (dict). Returns true.

create — Create a record

pygisceclient ... --model res.partner --method create \
  --args '[{"name": "ACME Corp", "customer": true}]'

Returns the new record's integer ID.

unlink — Delete records

pygisceclient ... --model res.partner --method unlink \
  --args '[[42, 43]]'

name_get — Human-readable names

pygisceclient ... --model res.partner --method name_get \
  --args '[[1, 2, 3]]'

Returns [[id, "display name"], ...].

name_search — Autocomplete search

pygisceclient ... --model res.partner --method name_search \
  --args '["Acme"]' \
  --kwargs '{"operator": "ilike", "limit": 10}'

exists — Check if a record exists

pygisceclient ... --model res.partner --method exists \
  --args '[42]'

Custom business methods

Models may expose methods beyond CRUD. Call them the same way:

pygisceclient ... --model account.invoice --method invoice_open \
  --args '[[42]]'

---

4. MsgPack Services

When using https+msgpack:// or http+msgpack://, the CLI can call MsgPack services directly with --service.

pygisceclient \
  --url https+msgpack://erp.example.com \
  --database mydb \
  --token "$ERP_TOKEN" \
  --service common \
  --method check_for_features \
  --args '[["feature_a", "feature_b"]]'

Service calls use positional JSON arguments only; --kwargs is reserved for model calls. The default --service-auth auto mode calls common, db and wc raw, because many of their methods do not receive database, uid and password. Other services are called authenticated.

Force the mode only when the ERP service expects the opposite convention:

pygisceclient ... --service report --service-auth authenticated \
  --method report --args '["account.invoice", [1], {}, {}]'

pygisceclient ... --service common --service-auth raw \
  --method check_for_features --args '[["webclient_feature"]]'

In Python, use MsgPackClient.service(name, authenticated=None) for the same behavior. None means auto mode; pass False for raw or True for authenticated:

features = client.common.check_for_features(['webclient_feature'])
databases = client.db.list()
report_id = client.service('report').report('account.invoice', [1], {}, {})

---

5. Domain Syntax (Search Filters)

Domains are lists of condition tuples following Polish (prefix) notation.

Basic tuple

("field_name", "operator", value)

Operators

Operator	Description
=, !=	Equals / Not equals
>, >=, <, <=	Comparisons
like, ilike	SQL LIKE (case-sensitive / insensitive)
not like, not ilike	Negated variants
in, not in	Value in / not in list
child_of	Hierarchical descendant

Logical operators (prefix)

• "&" — AND (default, implicit between tuples)
• "|" — OR
• "!" — NOT (unary)

Examples

Simple AND (implicit):

[["active", "=", true], ["city", "ilike", "barcelona"]]

OR:

["|", ["state", "=", "open"], ["state", "=", "draft"]]

Complex:

["&", "|", ["type", "=", "out_invoice"], ["type", "=", "out_refund"], ["state", "=", "open"]]

---

6. Field Types

Simple fields

Type	Description
char	Short text (VARCHAR)
text	Long text
integer / integer_big	Integer / Big integer
float	Decimal (NUMERIC)
boolean	True/false
date / datetime / time	Temporal types
binary	Binary data
selection	Enumeration
json / jsonb	JSON data

Relational fields

Type	Read returns	Write with
many2one	[id, "name"] or false	integer ID
one2many	list of IDs	special tuples (see below)
many2many	list of IDs	special tuples (see below)
reference	"model,id" string	"model,id" string

Writing relational fields (one2many / many2many)

Use special tuple commands in write/create vals:

Command	Meaning
[0, 0, {vals}]	Create new related record
[1, id, {vals}]	Update existing related record
[2, id, 0]	Delete related record
[3, id, 0]	Unlink (remove relation, keep record)
[4, id, 0]	Link an existing record
[5, 0, 0]	Unlink all
[6, 0, [ids]]	Replace all with given list

Functional / computed fields

• function: Computed by Python; may or may not be stored.
• related: Follows a chain of relations.
• property: Per-company configurable value.

---

7. Common Workflow Patterns

Discover a model's schema

pygisceclient ... --model res.partner --method fields_get | jq 'keys'

Search + Read

IDS=$(pygisceclient ... --model res.partner --method search \
  --args '[[["customer", "=", true]]]')

pygisceclient ... --model res.partner --method read \
  --args "[$IDS, [\"name\", \"vat\", \"city\"]]"

Paginated search

pygisceclient ... --model account.invoice --method search \
  --args '[[["state", "=", "open"]], 0, 100, "date_invoice desc"]'

Count before reading

pygisceclient ... --model res.partner --method search_count \
  --args '[[["active", "=", true]]]'

Discover all installed models

MODEL_IDS=$(pygisceclient ... --model ir.model --method search --args '[[]]')
pygisceclient ... --model ir.model --method read \
  --args "[$MODEL_IDS, [\"model\", \"name\"]]"

Discover fields via ir.model.fields

FIELD_IDS=$(pygisceclient ... --model ir.model.fields --method search \
  --args '[[["model", "=", "res.partner"]]]')
pygisceclient ... --model ir.model.fields --method read \
  --args "[$FIELD_IDS, [\"name\", \"field_description\", \"ttype\", \"relation\", \"required\", \"readonly\"]]"

---

8. Core Base Models

These models are always available:

Model	Description
res.partner	Contacts, companies, addresses
res.users	System users
res.company	Companies
res.country	Countries
res.country.state	States / provinces
res.currency	Currencies
ir.model	Model registry (all installed models)
ir.model.fields	Field definitions for all models
ir.model.access	Access control entries
ir.rule	Record-level security rules
ir.sequence	Sequence generators
ir.attachment	File attachments
ir.cron	Scheduled actions
account.invoice	Invoices
account.invoice.line	Invoice lines
account.account	Chart of accounts
product.product	Products
sale.order	Sale orders
purchase.order	Purchase orders
stock.picking	Stock movements

---

9. Best Practices

1. Always discover the schema first — call fields_get on a model before reading or writing.
2. Paginate large results — use offset and limit in search. Process in batches of 1000.
3. Request only needed fields — pass an explicit field list to read.
4. Use search_count before large reads to know how many records match.
5. Relational field values — many2one returns [id, "name"] on read; write with just the integer ID. one2many/many2many return lists of IDs.
6. JSON escaping — --args and --kwargs expect valid JSON. Use single quotes around JSON to avoid shell escaping issues.
7. Error handling — non-zero exit codes mean failure; error message is on stderr.
8. Security — the ERP enforces access control. Operations may fail with AccessError if the user lacks permissions.