terraform-provider-stackgen

sg_diary_insights (Data Source)

⚠️ Experimental — Lists Guild diary learning insights (suggested improvements extracted from agent diaries and episodic memory).

Typical uses:

• Read status = "proposed" suggestions into reporting or ticket automation
• Export approved / materialized insights (titles, categories, proposed actions) into docs or locals
• Correlate insights with agents via agent_name / category filters

This data source is read-only. Approving or dismissing insights is done in the Guild UI or Insights API, not Terraform. For raw daily diary text use sg_agent_diaries. Requires Insights to be configured (API may return 503 otherwise). Respects provider workspace_id for workspace scoping.

Example Usage

# When to use: Export proposed diary learning insights into tickets/reporting.
# Approving/dismissing insights is done in the Guild UI — this DS is read-only.
data "sg_diary_insights" "proposed" {
  status = "proposed"
  limit  = 25
}

output "proposed_insight_titles" {
  value = [for i in data.sg_diary_insights.proposed.insights : i.title]
}

Schema

Optional

• agent_name (String) Filter insights associated with this agent name.
• category (String) Filter by category (recurring_failure, task_sequence, policy_friction, efficiency_opportunity, missing_capability, cost_pattern).
• date (String) Filter by diary date (YYYY-MM-DD).
• limit (Number) Maximum number of insights to return.
• offset (Number) Number of insights to skip for pagination.
• status (String) Filter by review status: proposed, approved, dismissed, or materialized. Use proposed to list pending suggestions.
• tag (String) Filter by source tag (Guild defaults to source:diary-learning when omitted).

Read-Only

• insights (Attributes List) Diary learning insights (suggestions) matching the query. (see below for nested schema)
• total (Number) Total matching insights reported by the API (may exceed the returned page).

Nested Schema for insights

Read-Only:

• agent_names (List of String) Agents whose diaries contributed to this insight.
• category (String) Insight category (recurring_failure, task_sequence, policy_friction, efficiency_opportunity, missing_capability, cost_pattern).
• confidence (Number) Model confidence score for this insight.
• created_at (String) Creation timestamp (RFC 3339).
• date_range (String) Diary date range covered by this insight.
• description (String) Longer explanation of the detected pattern.
• evidence (Attributes) Signals that support this insight. (see below for nested schema)
• id (String) Unique diary insight UUID.
• proposed_action (Attributes) Suggested follow-up action (e.g. create a skill or runbook). (see below for nested schema)
• reviewed_at (String) When the insight was reviewed (RFC 3339).
• reviewed_by (String) Operator who approved or dismissed the insight.
• status (String) Review lifecycle status (proposed, approved, dismissed, materialized).
• tags (List of String) Tags associated with the insight.
• title (String) Short human-readable insight title.
• updated_at (String) Last update timestamp (RFC 3339).

Nested Schema for insights.evidence

Read-Only:

• audit_event_types (List of String) Audit event types cited as evidence.
• diary_entry_agent (String) Agent name for the linked diary entry.
• diary_entry_date (String) Diary entry date (YYYY-MM-DD) linked to this insight.
• episodic_memory_ids (List of String) Episodic memory point IDs cited as evidence.

Nested Schema for insights.proposed_action

Read-Only:

• materialized_ref (Attributes) Link to a draft artifact created after approval, when present. (see below for nested schema)
• rationale (String) Why this action was proposed.
• target_name (String) Suggested target name for the action (e.g. skill or runbook name).
• type (String) Proposed action type.

Nested Schema for insights.proposed_action.materialized_ref

Read-Only:

• href (String) Optional UI or API href for the artifact.
• id (String) Materialized artifact identifier.
• name (String) Materialized artifact name.
• status (String) Materialized artifact status.
• type (String) Materialized artifact type (e.g. skill, runbook).
• version (Number) Materialized artifact version when applicable.