Resources and filters
MCP adoption tools require one company_id or ticker and an approved user grant.
Sources are npm, pypi, nuget, docker, and terraform. Metadata resources accept project_id and source; use page_size (1–500, default 100) and cursor for continuation.
For agent access, use MCP: get_adoption_series defaults to grain=weekly and series_kind=aggregate; select series_kind=versions for version observations.
Activity fields
REST, MCP, and the logical delivery tables share activity-point fields.
| Field | Meaning |
|---|---|
| point_id | Stable activity-point identity shared with the logical warehouse projection. |
| company_id / project_id / artifact_id | Canonical company and published project/artifact identifiers. Do not substitute frontend project keys. |
| source / entity_id / metric | Source ecosystem, exact source artifact identity, and activity measure. |
| period_start / period_end | ISO dates. Daily endpoints describe one day; weekly endpoints describe a Monday–Sunday interval. Filters select period_end. |
| value | Cleaned activity as a JSON number, or null for a missing aggregate observation. Fractional values can occur. |
| adjusted_value | Aggregate activity after published event adjustment; falls back to value where no adjustment differs. |
| version_id / version / kind | Version-series fields: stable version ID, exact version label, and named/other/unattributed kind. Aggregate rows omit them. |
Logical tables are ADOPTION_DAILY, ADOPTION_WEEKLY, ADOPTION_VERSIONS_DAILY, and ADOPTION_VERSIONS_WEEKLY. S3 and Snowflake delivery are available by arrangement; this reference does not provision a delivery destination.
Metadata fields
Resolve dictionaries before interpreting or joining observations.
| Resource | Public fields |
|---|---|
| Projects | project_id, company_id, ticker, project_key, name, taxonomy, ownership, ai_profile |
| Artifacts | artifact_id, company_id, project_id, source, entity_id, metric |
| Versions | version_id, company_id, artifact_id, version, kind, first_observed, last_observed |
| Events | event_id, company_id, project_id, date, subject_key, source, metric, event_type, effect_type, maturity, direction, impact_percent, versions, label |
Coverage and pagination
Availability and publication identity are part of a reproducible extract.
Daily retention is up to 730 calendar days. All available complete aggregate weeks are calculated before daily truncation, so weekly history can be longer. Source coverage can be shorter; absent history is never fabricated.
Aggregate gaps remain null. Version membership is sparse and bounded upstream: an absent version point does not establish zero usage. Residual and unattributed version members remain visible. Version rows do not carry aggregate event adjustments.
Series sort by period_end, artifact_id, and version_id. Metadata sorts by its canonical identifier. Follow meta.next_cursor with the same filters; changing filters, exceeding cursor expiry, or changing the underlying publication requires restarting the extract.
Every response includes data_schema_version, publication_id, data_updated_at, row count, and pagination metadata. Invalid parameters return 422; unpublished company coverage returns 404; missing entitlement returns 403.
Read the Adoption Signals methodology and interpretation boundaries.