Reference

Everything here matches the current release. The README on GitHub has the longer explanations.

Commands

Command	What it does
`rendercache serve`	Runs the public proxy and the separate admin listener, warms sitemap pages and serves saved copies to crawlers
`rendercache render --url /path`	Renders one page in Chrome now and saves it
`rendercache inspect --url /path`	Prints the saved record for a page without starting Chrome
`rendercache prune --older-than 720h`	Removes old records while the service is stopped
`rendercache health`	Checks the admin health endpoint, using `RENDERCACHE_ADMIN_TOKEN`
`rendercache version`	Prints the version, commit and platform

Each command takes --config file.json. Common settings also have flags, such as --origin, --public-url, --listen, --admin-listen, --data-dir, --chrome-path, --ready-selector, --render-timeout, --warm, --sitemap, --snapshots and --snapshot-ttl. Flags override the file.

Settings

The config file is one JSON object. Unknown keys are rejected. RENDERCACHE_ADMIN_TOKEN and RENDERCACHE_EDGE_TOKEN override the token settings, so secrets can stay out of the file.

Setting	Default	What it does
`origin`	required	Fixed upstream site root
`public_url`	origin	Public root used for URL validation and snapshot identity
`data_dir`	`data/snapshots`	Local snapshot and delivery-log directory, one writer process
`snapshot_max_bytes`	`1073741824`	Serialized-record quota; 4096 bytes to 1 TiB
`snapshot_max_files`	`10000`	Record count quota; 1 to 100000
`max_request_bytes`	`16777216`	Public request-body limit; 1 byte to 256 MiB
`http_concurrency`	`32`	Concurrent public requests (1..256); excess requests return 503
`edge_token`	empty	Optional public-listener edge key; prefer environment
`delivery_log_max_bytes`	`10485760`	Maximum bytes per log file; 4096 to 1 GiB
`delivery_log_backups`	`3`	Rotated files retained in addition to current file; 1 to 10
`chrome_path`	auto-detect	Local browser executable
`chrome_no_sandbox`	`false`	Explicit browser sandbox opt-out
`render_timeout`	`30s`	Hard deadline including capacity wait/startup
`render_settle`	`500ms`	Wait after load when no selector is configured
`ready_selector`	empty	CSS selector proving application-specific readiness
`max_html_bytes`	8388608	Serialized HTML limit, at most 64 MiB
`render_concurrency`	1	Browser worker slots, range 1–16
`render_allow_origins`	empty	Additional explicit HTTP(S) asset roots
`snapshot_enabled`	`true`	Serve saved responses to eligible bots
`snapshot_ttl`	`24h`	HTTP freshness age, minimum 1s
`request_page_limit`	256	Request-driven page allowance with least-recently-requested eviction, range 1–10000
`warm_enabled`	`true`	Enable warming while serving
`sitemap_urls`	`["/sitemap.xml"]`	Public root sitemap URLs/paths
`warm_interval`	`24h`	Snapshot refresh age; minimum 1s
`sitemap_interval`	`15m`	Rediscovery schedule; minimum 1s
`warm_retry_base`	`5s`	Initial retry delay; minimum 100ms
`warm_retry_max`	`5m`	Maximum retry delay, at least base
`warm_queue_size`	128	Pending job capacity, range 1–10000
`sitemap_max_bytes`	8388608	Wire/XML bytes per document, at most 64 MiB
`sitemap_max_urls`	10000	Maximum discovered pages, at most 100000
`sitemap_max_documents`	128	Maximum sitemap documents, at most 1000
`sitemap_max_depth`	4	Maximum index depth, root at 0, at most 16
`sitemap_timeout`	`15s`	Timeout for one HTTP fetch
`sitemap_scan_timeout`	`2m`	Deadline for complete discovery
`listen`	`:8080`	Public proxy listener
`admin_listen`	`127.0.0.1:9090`	Administration listener
`admin_token`	empty	Optional bearer token on loopback, required elsewhere
`origin_timeout`	`30s`	Upstream response-header timeout
`shutdown_timeout`	`10s`	Graceful shutdown deadline
`bots`	search, AI and preview crawler patterns, including Google-InspectionTool for Search Console tests	Case-insensitive user-agent substring matches
`bypass_paths`	`["/api"]`	Path prefixes that always go to the origin, for example `/graphql` or `/wp-json`
`ignore_query_params`	utm_*, gclid, fbclid, msclkid	Query parameters left out of the snapshot identity, case-insensitive
`trusted_proxies`	`["127.0.0.0/8", "::1/128"]`	Peers whose X-Forwarded-For/Proto/Host are passed to the origin

Admin endpoints

These live only on the admin listener (127.0.0.1:9090 by default), never on the public one. With a token set, send Authorization: Bearer <token> or sign in as rendercache with the token as password.

Endpoint	What it does
`GET /`	The “What the bot sees” page: coverage, errors, receipts and the HTML comparison
`GET /healthz`	Process health
`GET /status`	Warming coverage, queue, cache counters and recent delivery receipts as JSON
`POST /purge?url=/path`	Queues a fresh render of one page and keeps the old copy until it succeeds. Always needs a token
`POST /diagnose`	Compares the raw origin HTML of `{"url":"/path"}` with the saved copy
`POST /simulate`	Sends a local crawler-style request through the public handler and records a receipt marked simulated

Response header

Every public response carries X-RenderCache:

Value	Meaning
`HIT`	A fresh saved copy was served to a crawler
`STALE`	An older saved copy was served while a refresh runs in the background
`MISS`	A crawler asked for a page with no saved copy yet; the origin answered and a render was queued
`BYPASS`	Not eligible for a saved copy: a person, a request with cookies or credentials, an asset, a bypassed path or a non-GET request