records
Introspecting THIS Cerb install's record types with the `cerb records` command line -- aliases, searchable filters, writable fields, and why those last two are separate namespaces. Load it before naming any field or filter.
Install
npx skills add https://github.com/jstanden/cerb/tree/12.0.x/features/cerberusweb.core/assets/agent_filesystems/cerb-agents/skills/records
claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install jstanden-cerb@llmmart
git clone https://github.com/jstanden/cerb.git
The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole jstanden/cerb collection as a plugin from our marketplace. Git is the plain clone.
Skill manifest
Record types, filters, and fields
The cerb command line — this install, not the generic baseline
Your terminal tool carries more than files: cerb is a command line into THIS Cerb install, so you can confirm a record type's alias and its real keys instead of guessing. Prefer it over the docs for lookups — it's one command, and it reflects this installation, including its custom record types and custom fieldsets. The docs describe the generic baseline: they can omit what someone is actually looking at, and list types a plugin here doesn't provide.
cerb records <sub> help prints usage; rows pipe like any command.
cerb records types [--filter <text>]— every record type, with thealiasthatrecord.*, a data query'sof:, andlinks.<alias>:all take.cerb records types --filter task→task,task_project.cerb records filters <type>[,<type>...]— the keys you can SEARCH by.cerb records filters ticketreturns akey/typetable:subjectText,importanceNumber,createdDate,owner.idWorker,group.idRecord,statusVirtual,messages.count.inNumber,spam.scoreDecimal,links.<record-type>Record, plus this install's custom fieldset keys (e.g.testFieldset.date).cerb records fields <type>[,<type>...]— the keys you can WRITE, i.e. whatrecord.create/record.updateaccept, with a required marker and notes. These are NOT query filters; reach for it when someone asks what a field is called on the record itself.
Filters and fields are separate namespaces
Never lift a key from fields into a query. On task the writable field is owner_id but the search filter is owner.id; writable status_id takes 0/1/2 while the status filter takes open/waiting/closed.
The type: column tells you which syntax applies
Text, Number, Decimal, Date, Seconds, Yes/No, Geo, Worker, Record (a chooser — numeric ids), Search (deep search), Virtual (its own vocabulary, e.g. status:open, watchers:me). The search-queries skill has one expression form per type.
Pipe rather than eyeball a long table
cerb records filters ticket | filters|filter(r => r.type == "Date")|column("key")
cerb records types | types|filter(r => r.is_custom)|column("alias")
Docs as fallback
cerb records is the authority for keys. Fall back to cerb-docs for the things it doesn't cover — background and semantics — and cite what you relied on:
/cerb-docs/references/docs/records/types.md— the baseline record type list (the index), plustypes/<type>.mdfor one type's background (e.g.types/ticket.md). Each type page lists its Records API fields, dictionary placeholders, Search Query Fields, and worklist columns./cerb-docs/references/docs/records/fields/types/— what a field TYPE means:text,number,float,timestamp,boolean,context,links,url,object,image,extension.
Files (cerb)
-
SKILL.md 3.1 KB
--- name: records description: Introspecting THIS Cerb install's record types with the `cerb records` command line -- aliases, searchable filters, writable fields, and why those last two are separate namespaces. Load it before naming any field or filter. --- # Record types, filters, and fields ## The `cerb` command line — this install, not the generic baseline Your terminal tool carries more than files: `cerb` is a command line into THIS Cerb install, so you can confirm a record type's alias and its real keys instead of guessing. Prefer it over the docs for lookups — it's one command, and it reflects this installation, including its custom record types and custom fieldsets. The docs describe the generic baseline: they can omit what someone is actually looking at, and list types a plugin here doesn't provide. `cerb records <sub> help` prints usage; rows pipe like any command. - `cerb records types [--filter <text>]` — every record type, with the `alias` that `record.*`, a data query's `of:`, and `links.<alias>:` all take. `cerb records types --filter task` → `task`, `task_project`. - `cerb records filters <type>[,<type>...]` — **the keys you can SEARCH by.** `cerb records filters ticket` returns a `key`/`type` table: `subject` Text, `importance` Number, `created` Date, `owner.id` Worker, `group.id` Record, `status` Virtual, `messages.count.in` Number, `spam.score` Decimal, `links.<record-type>` Record, plus this install's custom fieldset keys (e.g. `testFieldset.date`). - `cerb records fields <type>[,<type>...]` — the keys you can WRITE, i.e. what `record.create`/`record.update` accept, with a required marker and notes. These are NOT query filters; reach for it when someone asks what a field is called on the record itself. ## Filters and fields are separate namespaces **Never lift a key from `fields` into a query.** On `task` the writable field is `owner_id` but the search filter is `owner.id`; writable `status_id` takes `0`/`1`/`2` while the `status` filter takes `open`/`waiting`/`closed`. ## The `type:` column tells you which syntax applies Text, Number, Decimal, Date, Seconds, Yes/No, Geo, Worker, Record (a chooser — numeric ids), Search (deep search), Virtual (its own vocabulary, e.g. `status:open`, `watchers:me`). The `search-queries` skill has one expression form per type. ## Pipe rather than eyeball a long table ``` cerb records filters ticket | filters|filter(r => r.type == "Date")|column("key") cerb records types | types|filter(r => r.is_custom)|column("alias") ``` ## Docs as fallback `cerb records` is the authority for keys. Fall back to `cerb-docs` for the things it doesn't cover — background and semantics — and cite what you relied on: - `/cerb-docs/references/docs/records/types.md` — the baseline record type list (the index), plus `types/<type>.md` for one type's background (e.g. `types/ticket.md`). Each type page lists its Records API fields, dictionary placeholders, Search Query Fields, and worklist columns. - `/cerb-docs/references/docs/records/fields/types/` — what a field TYPE means: `text`, `number`, `float`, `timestamp`, `boolean`, `context`, `links`, `url`, `object`, `image`, `extension`.
Comments (0)
Sign in to join the conversation.
Reviews (0)
No reviews yet.
No comments yet.