@thinkingai/ae-cli 6.1.12 → 6.1.14

package/skills/ae-analysis/references/catalog_export.md

@@ -0,0 +1,28 @@
+# analysis-meta catalog export
+
+## Use
+
+Export the complete permission-filtered analysis metadata catalog for conversation-local reuse.
+
+Do not use it for ordinary discovery, bounded search, or when compiler candidates already exist; use `catalog list` for the one aggregate online search.
+
+```bash
+ae-cli analysis-meta catalog export \
+  --project-id <project_id> \
+  --output <conversation_catalog_dir>/catalog.jsonl
+```
+
+Capability id: `metadata.catalog.export`.
+
+## Input
+
+The gateway input contains only `project_id`. The response must contain every accessible event, metric, event property, user property, cluster, and tag row with `complete=true` and `total` equal to the row count.
+
+## Output
+
+The CLI validates completeness, publishes the JSONL atomically with mode `0600`, and writes the adjacent `catalog.meta.json` last. Search the file locally and keep full rows out of model context. Reuse a valid snapshot only for the same host, project, principal, and conversation.
+
+| Parameter | Required | Description |
+|---|---|---|
+| `--project-id` | Yes | Numeric project ID. |
+| `--output` | Yes | Conversation-local `.jsonl` path. |

package/skills/ae-analysis/references/catalog_list.md

@@ -0,0 +1,44 @@
+# analysis-meta catalog list
+
+Use only for a structured AI-QP metadata compile failure that requires one aggregate online search across selected resource types.
+
+Do not use it for ordinary asset discovery or when the compiler already returned candidates. Follow [`../metadata_resolution.md`](../metadata_resolution.md).
+
+Online search:
+
+```bash
+ae-cli analysis-meta catalog list \
+  --project-id <project_id> \
+  --queries '["付费事件","支付事件","充值事件"]' \
+  --resource-types '["event","metric"]' \
+  --limit-per-type 20
+```
+
+Capability id: `metadata.catalog.list`.
+
+## Input
+
+Sends `project_id`, `queries`, `resource_types`, and optional `limit_per_type`. Queries are OR matched only inside the selected resource types.
+
+## Output
+
+Both modes return unified rows with:
+
+- `resource_type`: `event`, `metric`, `event_property`, `user_property`, `cluster`, or `tag`
+- `resource_key`: canonical server-defined identifier
+- `display_name`
+- `remark`
+- `scope`: `event` or `user` for property rows
+- property type fields when applicable
+
+Online rows also contain `matched_query`, `matched_field`, and `match_type`. Results are limited independently per resource type. `has_more=true` and `truncated_resource_types` identify types with additional matches; do not page them in the structured resolution workflow.
+
+## Parameters
+
+| Parameter | Required | Description |
+|---|---|---|
+| `--project-id` | Yes | Numeric project ID. |
+| `--queries` | Online mode | JSON array of 1–20 deduplicated keywords, OR matched. |
+| `--resource-types` | Online mode | JSON array containing only `event`, `metric`, `event_property`, `user_property`, `cluster`, or `tag`. |
+| `--limit-per-type` | No | Online result limit per selected resource type; default 20, maximum 200. |
+Use at most one online search for the complete compiler error array. If every path gets candidates, ask for confirmation. If any path remains empty, use [`catalog_export.md`](catalog_export.md) once.