Bundled specs
The npm package ships providers/—optional on-disk API descriptions so cold start avoids downloading multi‑MB specs. Most native GraphQL / gRPC upstreams are configured only via CLAWQL_GRAPHQL_SOURCES / CLAWQL_GRPC_SOURCES (Spec configuration). Exception: Linear is a bundled GraphQL-only vendor (providers/linear/schema.graphql); use CLAWQL_PROVIDER=linear or include linear in CLAWQL_BUNDLED_PROVIDERS / all-providers. For multi-spec, the only unconfigured default is the full merge all-providers. Any other multi-spec load is explicit: set CLAWQL_BUNDLED_PROVIDERS, CLAWQL_SPEC_PATHS, or a merged CLAWQL_PROVIDER (for example google, atlassian, or all-providers set on purpose). See Spec configuration.
Learn walkthrough (Tika → Gotenberg → Stirling → Paperless → Onyx, orchestration): Document pipeline.
Canonical REST-forward matrix (9 vendors)
Use this as the default-bundle reference for REST / OpenAPI–heavy workflows. It matches the historical “nine vendor” docs story; google is a merged Discovery preset, not a single file.
Provider id (CLAWQL_PROVIDER / CLAWQL_BUNDLED_PROVIDERS) | Primary auth env(s) | Typical use in workflows |
|---|---|---|
cloudflare | CLAWQL_CLOUDFLARE_API_TOKEN | Edge and DNS operations, Workers deploy/debug paths, docs-site operations |
github | CLAWQL_GITHUB_TOKEN | Issues/PR/repo automation and release workflow actions |
slack | CLAWQL_SLACK_TOKEN / SLACK_BOT_TOKEN | Team notifications and Slack API operations; powers optional notify |
paperless | PAPERLESS_API_TOKEN (+ PAPERLESS_BASE_URL) | Document archive lookup and ingestion workflows |
stirling | STIRLING_API_KEY (+ STIRLING_BASE_URL) | PDF conversion/splitting/merge operations |
tika | CLAWQL_BEARER_TOKEN (+ TIKA_BASE_URL) | Text/metadata extraction from many file types |
gotenberg | CLAWQL_BEARER_TOKEN (+ GOTENBERG_BASE_URL) | HTML/office/markdown to PDF conversion |
onyx | ONYX_API_TOKEN (+ ONYX_BASE_URL) | Enterprise knowledge search via optional knowledge_search_onyx; optional ingestion via execute |
google | CLAWQL_GOOGLE_ACCESS_TOKEN / GOOGLE_ACCESS_TOKEN | Bundled Google Cloud Discovery merge (multiple APIs); part of default cross-provider search and execute flows |
Bundled GraphQL-only: linear
| Provider id | Primary auth env(s) | Notes |
|---|---|---|
linear | LINEAR_API_KEY / CLAWQL_LINEAR_API_KEY | GraphQL at https://api.linear.app/graphql. SDL vendored from Linear’s open-source SDK (providers/linear/schema.graphql). Refresh: npm run fetch-linear-schema. |
Presets and additional bundled ids
| Selection | Notes |
|---|---|
CLAWQL_BUNDLED_PROVIDERS= a,b,… | Explicit custom merge by id (including optional google) |
all-providers | Unconfigured default merged preset: Google + bundled vendors |
atlassian | Merged Jira + Bitbucket |
sentry, n8n, jira, bitbucket, linear | Additional single bundled provider ids (linear is GraphQL-only; see above) |
Aliases: jira / bitbucket map to their standalone bundles; atlassian merges both. Deprecated merged id: google-top50 -> google.
Flink connector sync and full-stack deployment topology are intentionally documented separately from the provider matrix: #119 (Flink connector pipeline) and Helm + #113 (full-stack deployment pointers).
Offline and refresh
Bundled files are read from disk first. If a file is missing, the registry may fetch a fallback URL unless CLAWQL_BUNDLED_OFFLINE=1.
Maintainers refresh specs with npm run fetch-provider-specs (see providers/README.md). Optional GraphQL artifacts: npm run pregenerate-graphql (Bun recommended).
Default first run
With no spec-related environment variables, ClawQL loads all-providers so cross-vendor search works immediately after install.