Munin
Munin developer portal
Get a key →
Section · Skills Library

How your agent works.

Markdown procedures the agent can read at runtime, written by humans who know the domain.

analytics 3

cms 7

skill://cms/author-with-blocks

CMS: Author content with blocks

Define typed block types on a collection and author rich in-article content (callouts, galleries, product cards) whose embedded assets and entry references are validated, expanded on read, and tracked for deletion safety.

863 words
skill://cms/design-collection

CMS: Design a collection

Turn "I need a blog / product catalog / FAQ / landing-page content store" into a working `cms_create_collection` call — pick field types deliberately, order fields for the editor, decide localization up front, and avoid the field-migration paper cuts.

1,980 words
skill://cms/localize-entry

CMS: Localize an entry

Configure locales, author entries per language, and switch the org's default locale without breaking already-published entries.

589 words
skill://cms/migrate-content

CMS: Migrate content in bulk

Move entries from an external CMS (or between Munin collections) idempotently — preserve references, rewrite assets, reconcile schema drift.

698 words
skill://cms/publish-entry

CMS: Publish an entry

Move a CMS entry through draft → published — immediate, scheduled, or rolled back — without losing work to optimistic-lock conflicts.

553 words
skill://cms/review-stale-entries

CMS: Review stale entries

Periodic curator pass — find drafts that have stalled, published entries that haven't been touched in months, and orphaned assets. Reports findings with recommended actions. No persistence layer; the operator reviews the curator-runner's log/reply and acts manually via existing CMS tools.

1,521 words
skill://cms/upload-asset-and-embed

CMS: Upload an asset and embed it

Two-phase asset upload (request presigned upload → send binary → complete), embed in entries, and audit unused assets.

867 words

conv 6

crm 6

skill://crm/clean-contact-data

CRM: Clean up contact data

Periodic curator pass — find duplicate / inconsistent contacts, file high-confidence pairs as structured merge proposals via crm_propose_merge. Designed to be run on a cadence by an admin agent (weekly is a good default). The operator reviews via crm_list_merge_proposals and resolves with crm_apply_merge_proposal / crm_dismiss_merge_proposal.

1,332 words
skill://crm/deduplicate-contacts

CRM: Deduplicate contacts

Find duplicate contacts (same person, multiple rows) and consolidate them. There is no merge tool — this skill documents the manual reconcile pattern.

674 words
skill://crm/extract-contact-from-message

CRM: Extract a contact from a message

When a conversation closes, read the thread for identifying info the end-user volunteered (name, email, phone, company, title) and persist it to the CRM as a contact — auto-applied, no proposal queue. Designed to fire on every `conversation.closed` event so visitor-volunteered identity becomes structured CRM data without operator intervention.

1,420 words
skill://crm/import-and-score-leads

CRM: Import and score leads

Bulk-import contacts, attach them to companies and a deal pipeline, log the source touchpoint, and seed AI summaries for downstream prioritization.

677 words
skill://crm/onboard-new-customer

CRM: Onboard a new customer

Recommended sequence for capturing a new customer, attaching them to a company, and seeding the AI summary that drives later prioritization.

496 words
skill://crm/progress-deal-through-pipeline

CRM: Progress a deal through the pipeline

Move a deal through pipeline stages with the right activity log at each gate, then refresh AI summary so prioritization stays fresh.

708 words

kb 4

outreach 3

playbooks 5

webhooks 1