๐ฎ๐ฉ Indonesia Government APIs & Data Sources
59 documented source records: 58 Indonesia-related records and 1 international reference. Includes government and non-government publishers, overlapping services and historical endpoints โ not 59 verified open APIs. 11 supplementary guides are counted separately.
The machine-readable catalog is the source inventory. Every entry records publisher, tier, documentation, portal URL, government/non-government classification, access notes, review state and dated evidence. Review findings and primary sources distinguish documentation confirmation, bounded endpoint observations and unverified history.
What changed in the October 10, 2026 review
- Filled all five previously undocumented tier-8 additions; included BAPPEBTI and Japan NTA explicitly. Tier 7 now contains seven records, including the preserved Aturan.org contribution.
- Withdrawn the BPJPH supervisor-as-certified-business example and totals. SIKePO is banking-regulation search, not a fintech/crypto registry; see the cited source docs.
- Removed blanket CKAN/API-works claims, speculative quotas and access-control bypass advice. Historical notes are marked superseded, not presented as current integrations.
- Public pages, account-gated tax/health systems and third-party indexes are explicitly different access classes. No credentials, personal-record queries or authenticated production integrations were used in this review.
- Issue #1 and PR #9 were reviewed read-only. PR #9 subsequently merged on October 10, 2026 (
21cfe4b); its Aturan.org source is preserved with documentation-only authentication/provenance caveats, not historical availability badges.
Monitoring is a separate inventory
The static status page, latest observation and monitor guide describe HTTP/transport observations, not API usability, licensing or record correctness. Catalog monitor_ids may map multiple endpoint probes to one source document. An empty array means unmonitored: BAPPEBTI, NTA, OGP Indonesia and Aturan.org. The preserved PR #10 expansion brings monitoring to 75 endpoints: 57 mapped to source documents and 18 monitoring-only entries, also recorded in their separate registry. These added probes do not inflate the 59-source documentation count. Monitor totals are validated against status/check.py, not inferred from documentation totals. Content markers are limited heuristics, not API/schema certification.
An absent regional observation is unknown, not evidence of a geo-block. One 403 or timeout does not prove nationwide or foreign-IP restrictions. Retained March 29, 2026 status notes are historical and internally inconsistent; do not reuse their totals or global availability claims as current evidence.
Link-status audit
The Markdown link audit and machine-readable audit snapshot record bounded, dated URL observations. They do not validate API schemas, authentication, data correctness or reuse permissions. Unreachable/blocked links remain explicitly unverified; an HTTP 200 landing page is not a working API.
Source index
Tier names are retained for navigation compatibility, not graded promises of availability. Unverified means inventory/document review only. Primary documentation reviewed confirms only the stated publisher documentation; endpoint observed is scoped to the recorded request, not all interfaces.
Tier 1 โ historically open-API candidates (12 records)
Tier 2 โ public web/search candidates (10 records)
Tier 3 โ regional data portals (6 records)
Tier 4 โ ministry-specific references (8 records)
Tier 5 โ transparency references (5 records)
Tier 6 โ financial references (4 records)
Tier 7 โ civic/legal/geospatial references (7 records)
Tier 8 โ formerly missing additions (5 records)
Additional Indonesia reference (unmonitored) (1 records)
International reference (unmonitored) (1 records)
Getting started safely
- Read the selected source's reviewed guidance and evidence, not its historical examples first.
- Prefer publisher documentation and licensed aggregate downloads. Obtain required credentials through official channels; never commit or print them.
- Verify endpoint, HTTP status, content type and response shape with a bounded request. Tests here are offline and do not establish live availability.
- Respect publisher limits and attribution. Stop at login, CAPTCHA, access denial or explicit restrictions; request approved access instead of routing around controls.
- Treat certification, registration, sector licensing and negative alerts as separate evidence. No result is not proof of compliance or legality.
For BMKG's official forecast interface and BPS key/response handling, see BMKG and BPS. The examples have independent safeguards and tests; consult their CLI help before running network requests.
AI-agent and MCP references
SKILL.md routes queries to canonical reviewed documentation. MCP guide separates third-party transport candidates from verified integrations. Never infer /tools/<name> REST routes from MCP tool names or allow a remote service to receive private records merely because it is publicly reachable.
Offline verification
python scripts/validate_catalog.py
python -m unittest discover -s tests -p 'test_catalog.py'
python scripts/check_repository.py
The catalog validator checks schema, exact source/reference coverage, computed totals, monitor mappings, README index coverage and local Markdown links/anchors. External URL checks are syntactic, not live liveness tests. See contribution guidance for the broader gates.
Related projects
- indonesia-civic-stack: separate SDK/MCP project; integration coverage is not verified by this catalog.
- indonesia-civic-signal-monitor: separate civic monitoring project; public reference unavailable (HTTP 404 observed October 10, 2026).
Disclaimer and license
Independent educational/research documentation, not affiliated with the listed publishers. Public visibility does not imply bulk-access permission, unrestricted reuse or permission to publish personal data. Verify source terms, original records and applicable requirements. MIT applies to repository content, not automatically to external datasets.