Move current GRASP specification links from GitHub to GitWorkshop. Link audit output to its exact pinned specification commit. Use NIP-05 cloning for ngit-grasp and ngit.dev for the ngit homepage. Leave the archived migration link unchanged as a historical record. Validated with rustfmt and the 54-test grasp-audit library suite.
58 KiB
Reference: Configuration
Purpose: Complete reference for all ngit-grasp configuration options
Audience: Operators and developers
Configuration Methods
ngit-grasp can be configured via:
- Environment variables (recommended for deployment)
.envfile (recommended for development)- Command-line arguments (planned, not yet implemented)
Configuration is loaded at startup and validated before the server starts.
Environment Variables
Server Configuration
NGIT_BIND_ADDRESS
Description: Address and port for the HTTP server to bind to
Type: String (IP:PORT format)
Default: 127.0.0.1:7334
Required: No
Examples:
# Localhost only (development)
NGIT_BIND_ADDRESS=127.0.0.1:7334
# All interfaces (production)
NGIT_BIND_ADDRESS=0.0.0.0:7334
# IPv6
NGIT_BIND_ADDRESS=[::1]:7334
# Custom port
NGIT_BIND_ADDRESS=127.0.0.1:3000
Notes:
- Use
127.0.0.1for local development - Prefer a loopback or private address behind a production reverse proxy
- Use
0.0.0.0only when the listener must be directly reachable - Ensure firewall rules allow the port
NGIT_TRUSTED_PROXY_CIDRS
Description: Comma-separated IPv4 or IPv6 networks whose forwarding headers may identify a WebSocket client
Type: CIDR list
Default: Empty
Required: No
Examples:
# Caddy or nginx connects over IPv4 loopback
NGIT_TRUSTED_PROXY_CIDRS=127.0.0.1/32
# Reverse proxy may connect over either loopback family
NGIT_TRUSTED_PROXY_CIDRS=127.0.0.1/32,::1/128
# Two private proxy tiers
NGIT_TRUSTED_PROXY_CIDRS=10.10.0.0/24,10.20.0.0/24
When the TCP peer is in one of these networks, ngit-grasp resolves the client
from X-Forwarded-For, Forwarded, or X-Real-IP, in that order. Forwarding
chains are walked from the TCP peer inward and stop at the first untrusted hop.
This resolved IP is used for relay connection policy, per-IP connection
metrics, abuse indicators, and WebSocket connection logs.
Forwarding headers are always ignored for untrusted TCP peers. Malformed headers also fall back to the peer address rather than a less authoritative header. Invalid CIDRs stop startup.
Security requirements:
- Leave this empty for a directly exposed ngit-grasp listener.
- Configure the reverse proxy to append or overwrite forwarding headers rather than passing client-controlled values unchanged.
- Bind ngit-grasp to loopback/private interfaces or firewall the backend so untrusted clients cannot connect from an address included in this list.
- Include every proxy hop that should be traversed. The first address outside the trusted ranges is treated as the client.
The corresponding NixOS option is trustedProxyCidrs.
NGIT_DOMAIN
Description: Public domain name for this GRASP instance
Type: String (domain name)
Default: None
Required: Yes
Examples:
NGIT_DOMAIN=gitnostr.com
NGIT_DOMAIN=git.example.org
NGIT_DOMAIN=localhost:7334 # Development only
Used for:
- NIP-11 relay information document
- Generating repository URLs
- CORS configuration
- Webhook URLs (future)
Notes:
- Must be accessible from the internet for production
- Include port if non-standard (e.g.,
localhost:7334) - Combined with
NGIT_BASE_PATHin repository clone and relay URLs
NGIT_BASE_PATH
Description: Public URL path where this GRASP instance is mounted
Type: String (normalized absolute URL path)
Default: /
Required: No
Examples:
NGIT_BASE_PATH=/
NGIT_BASE_PATH=/grasp
NGIT_BASE_PATH=/services/git
When set to a non-root path, every service endpoint is scoped below it. For
example, NGIT_DOMAIN=example.org with NGIT_BASE_PATH=/grasp exposes the
Nostr relay at wss://example.org/grasp, Git repositories below
https://example.org/grasp/<npub>/<repo>.git, and metrics at
https://example.org/grasp/metrics. Requests outside the prefix are rejected.
The value must start with /, must not end with / unless it is exactly /,
and must not contain empty, . or .. segments, a query, or a fragment.
Path-mounted relays do not serve or advertise the root-domain _@domain
NIP-05 identity. Their generated owner profile omits its nip05 field.
The corresponding NixOS option is basePath.
Nostr Relay Configuration
Relay owner identity
Description: Nostr secret key used for the relay operator identity
Type: String (nsec1... format)
Default: Load or create .relay-owner.nsec in the working directory
Required: No
Examples:
NGIT_RELAY_OWNER_NSEC=nsec1...
Used for:
- Deriving the
_@domainNIP-05 identity served from/.well-known/nostr.jsonwhen the relay itself is available at the domain root (/) - Signing a minimal kind-0 profile containing only the scheme-less public URL
as
name,nip05: "_@domain", andbot: true, plus a kind-10002 list naming this relay as its sole read/write relay; a kind with a stored local event keeps the stored version, and nothing is published toNGIT_USER_INDEX_RELAYSbefore at least one of them has been checked for the kind (and never whenNGIT_PRIVATE_MODEis enabled) - Trusted admission of valid events authored by this identity for kinds
without a dedicated admission policy, such as
ngit-cicoordinator and result events that do not reference a repository. Owner-signed NIP-34 repository events pass the normal admission policies, NIP-09/NIP-62 tombstones still apply, and because the event blacklist cannot block this key, rotating it is the only remediation if it is compromised - Deriving the operator pubkey in the NIP-11 relay information document
- NIP-42 authentication when synchronizing from other relays
Notes:
- The key is never accepted on the command line, where it would be exposed by process listings.
- Loading precedence is the systemd credential named
relay_owner_nsec,NGIT_RELAY_OWNER_NSEC(including.env), then.relay-owner.nsec. - NixOS operators should set
relayOwnerNsecFile; the module supplies that file as a protected systemd credential. - A configured credential or environment value that is empty or invalid stops startup. It never falls through to generating a replacement identity.
- If no source is configured and
.relay-owner.nsecdoes not exist, ngit-grasp generates it with mode0600.
NGIT_RELAY_NAME
Description: Human-readable name for this relay
Type: String
Default: "ngit-grasp relay"
Required: No
Examples:
NGIT_RELAY_NAME="GitNostr Community Relay"
NGIT_RELAY_NAME="Alice's GRASP Server"
Used for:
- NIP-11 relay information document
- Client display
- Relay discovery
NGIT_RELAY_DESCRIPTION
Description: Description of this relay's purpose and policies
Type: String
Default: "A GRASP-compliant Git relay"
Required: No
Examples:
NGIT_RELAY_DESCRIPTION="Public GRASP relay for open source projects"
NGIT_RELAY_DESCRIPTION="Private relay for ACME Corp repositories"
Used for:
- NIP-11 relay information document
- User information
- Relay selection
Storage Configuration
NGIT_GIT_DATA_PATH
Description: Directory path for storing Git repositories
Type: String (filesystem path)
Default: ./data/git
Required: No
Examples:
# Relative path (development)
NGIT_GIT_DATA_PATH=./data/git
# Absolute path (production)
NGIT_GIT_DATA_PATH=/var/lib/ngit-grasp/git
# Custom location
NGIT_GIT_DATA_PATH=/mnt/storage/git-repos
Storage structure:
{NGIT_GIT_DATA_PATH}/
├── {npub1}/
│ ├── {repo1}.git/
│ │ ├── objects/
│ │ ├── refs/
│ │ └── ...
│ └── {repo2}.git/
└── {npub2}/
└── ...
Notes:
- Directory must be writable by ngit-grasp process
- Ensure sufficient disk space
- Consider backup strategy
- Use fast storage for better performance
NGIT_RELAY_DATA_PATH
Description: Directory path for storing Nostr events and relay data
Type: String (filesystem path)
Default: ./data/relay
Required: No
Examples:
# Relative path (development)
NGIT_RELAY_DATA_PATH=./data/relay
# Absolute path (production)
NGIT_RELAY_DATA_PATH=/var/lib/ngit-grasp/relay
# Separate disk
NGIT_RELAY_DATA_PATH=/mnt/ssd/relay-data
Storage structure:
{NGIT_RELAY_DATA_PATH}/
├── events/
│ └── {event-id}.json
├── indexes/
│ ├── by-kind/
│ ├── by-author/
│ └── by-tag/
└── metadata/
Notes:
- Directory must be writable
- Consider SSD for better query performance
- Size grows with event count
- Implement retention policy for production
NGIT_DATABASE_BACKEND
Description: Database backend type for storing Nostr events
Type: String (enum: memory, lmdb)
Default: lmdb
Required: No
Valid Values:
lmdb- LMDB backend (persistent, general purpose)memory- In-memory database (fastest, no persistence)
Examples:
# Production (default, persistent)
NGIT_DATABASE_BACKEND=lmdb
# Development/testing (no persistence)
NGIT_DATABASE_BACKEND=memory
Comparison:
| Backend | Persistence | Performance | Use Case |
|---|---|---|---|
| lmdb | Yes | High | Production (general purpose) |
| memory | No | Fastest | Development, testing |
Notes:
memorybackend loses all data on restartlmdbbackend usesNGIT_RELAY_DATA_PATHfor storage- Default
memorybackend suitable for development and testing only - Production deployments should use
lmdb
NGIT_STARTUP_INTEGRITY_IDENTIFIERS
Description: Comma-separated repository identifiers to check during the automatic startup storage- and authorization-integrity passes
Type: String list
Default: Empty, meaning every installed identifier family
Required: No
# Staged validation of two families
NGIT_STARTUP_INTEGRITY_IDENTIFIERS=repo-one,repo-two
Use a non-empty value only to validate a release candidate on selected
families while the relay remains online. Remove the setting for the full
security sweep: scoped startup logs establish integrity only for the named
identifiers. Invalid or duplicate identifiers stop startup. The corresponding
NixOS option is startupIntegrityIdentifiers.
Proactive Sync Configuration (GRASP-02)
These options configure the proactive sync feature that synchronizes events from other relays.
NGIT_SYNC_PLUS_ENABLED
Description: Enable GRASP-03 Sync+ mailbox discovery on top of proactive GRASP-02 sync
Type: Boolean
Default: true
Required: No
# Opt out of mailbox discovery while retaining ordinary proactive sync
NGIT_SYNC_PLUS_ENABLED=false
The corresponding NixOS option is syncPlusEnabled. The setting is effective
only while the relay service's proactive sync manager is running. When false,
the relay retains GRASP-02 sync but does not discover NIP-65 inboxes and omits
GRASP-03 from its NIP-11 supported_grasps list.
NGIT_SYNC_RECURSIVE_DESCENDANT_LIMIT
Description: Soft limit on recursive query-frontier members below each event which directly tags a repository root
Type: Positive integer
Default: 500
Required: No
# Exercise branch saturation in a fresh archive or test environment
NGIT_SYNC_RECURSIVE_DESCENDANT_LIMIT=3
Direct repository-root events and events which independently tag a root do not consume this allowance. Once the deterministic breadth-first frontier contains the configured number of descendants for a branch, none of that branch's members are used in further child queries. Already active requests can still store more than the configured number, which is why this is a soft limit; those extra events do not extend the branch. Startup and historic reconciliation derive the same frontier from locally stored events, so a saturated branch does not receive a fresh budget. Source relays retain independent connections and sync progress; there is no cross-relay counter or coordinator.
The corresponding NixOS option is syncRecursiveDescendantLimit.
NGIT_SYNC_BOOTSTRAP_RELAY_URL
Description: URL of the bootstrap relay to initially sync events from Type: String (WebSocket URL) Default: None (relay discovery only) Required: No
Examples:
# Sync from a public relay
NGIT_SYNC_BOOTSTRAP_RELAY_URL=wss://relay.example.com
# Sync from another GRASP relay
NGIT_SYNC_BOOTSTRAP_RELAY_URL=wss://git.nostr.dev
# Local testing
NGIT_SYNC_BOOTSTRAP_RELAY_URL=ws://127.0.0.1:8081
Notes:
- Bootstrap relay provides initial sync source on startup
- Additional relays are automatically discovered from repository announcements that list our service
- Even without a bootstrap relay, sync will discover relays from stored announcements
- Synced events go through the same validation as directly-submitted events
- Use WebSocket protocol (
ws://orwss://) or defaults to wss:// - The bootstrap relay is operator-configured and therefore exempt from the
outbound target policy (see
NGIT_SYNC_ALLOW_NON_GLOBAL_TARGETS), so a local bootstrap relay keeps working even with the policy enforced
NGIT_USER_INDEX_RELAYS
Description: Comma-separated relay URLs receiving the relay-owner kind 0/10002 identity events and used to discover eligible accepted repository participants' NIP-65 relay lists
Type: String list (comma-separated WebSocket URLs)
Default: wss://purplepag.es,wss://index.hzrd149.com,wss://indexer.coracle.social
Required: No
NGIT_USER_INDEX_RELAYS=wss://purplepag.es,wss://index.example.com
The corresponding NixOS option is userIndexRelays, expressed as a list of
strings. Empty comma-separated entries and surrounding whitespace are ignored.
On startup, ngit-grasp publishes the owner's kind-0 and kind-10002 identity
events to the valid configured user-index relays, gap-filling only: no
identity event — locally stored or freshly generated — is published before
at least one user-index relay has been successfully checked for that kind,
and every send is preceded by a per-relay re-check. An identity found on an
index relay (for example after a local database wipe) is adopted locally and
never overwritten, so events reach only index relays that individually
confirm they hold no identity of that kind. For a kind with no local copy,
nothing is even seeded until a reachable user-index relay confirms it holds
none. Transient failures are retried with a capped backoff; terminal
protocol rejections are logged without retrying. Neither remote outages nor
local rejections block relay startup. With no user-index relays configured —
or when NGIT_PRIVATE_MODE is enabled — missing identity events are seeded
locally right away and published nowhere.
NGIT_SYNC_PLUS_FALLBACK_RELAYS
Description: Bounded inbox fallback relays for eligible Sync+ authors whose NIP-65 list was not found
Type: String list (comma-separated WebSocket URLs)
Default: wss://relay.ditto.pub,wss://relay.damus.io,wss://nos.lol,wss://relay.primal.net
Required: No
NGIT_SYNC_PLUS_FALLBACK_RELAYS=wss://relay.ditto.pub,wss://nos.lol
The corresponding NixOS option is syncPlusFallbackRelays. Fallback coverage
starts only after a successful user-index query returns no accepted kind
10002 for an eligible author. A subsequently discovered relay list replaces
the fallback in desired coverage without eagerly closing shared subscriptions.
Set the list empty to disable this recovery while retaining ordinary Sync+.
NGIT_SYNC_ALLOW_NON_GLOBAL_TARGETS
Description: Allow event-directed sync targets that are not globally reachable
Type: Boolean
Default: false
Required: No
Examples:
# Production (default): reject non-global event-directed targets
NGIT_SYNC_ALLOW_NON_GLOBAL_TARGETS=false
# Integration tests / closed development networks only
NGIT_SYNC_ALLOW_NON_GLOBAL_TARGETS=true
Notes:
- Repository announcements and PR events are untrusted; their
relaysandcloneURLs feed outbound WebSocket connections andgit fetchcalls - With the default (
false), event-directed targets are rejected when they use loopback, private, link-local, CGNAT, or other special-purpose addresses, local hostnames (localhost, single-label names,.local,.internal,.home.arpa,.onion, ...), embedded credentials, or hostnames that resolve to non-globally-reachable addresses (SSRF protection) - Scheme and credential checks still apply when set to
true - The operator-configured
NGIT_SYNC_BOOTSTRAP_RELAY_URLis always allowed, even when local; event-provided URLs never inherit that exception - Leave
falsein production;trueis intended for tests and closed development networks - See
docs/explanation/defensive-measures.mdfor the full outbound target policy, including the remaining DNS-rebinding caveat for relay connections
NGIT_SYNC_MAX_BACKOFF_SECS
Description: Maximum backoff time in seconds for sync relay reconnection
Type: Integer (seconds)
Default: 3600 (1 hour)
Required: No
Examples:
# Default: 1 hour max backoff
NGIT_SYNC_MAX_BACKOFF_SECS=3600
# Aggressive: 5 minute max backoff
NGIT_SYNC_MAX_BACKOFF_SECS=300
# Conservative: 2 hour max backoff
NGIT_SYNC_MAX_BACKOFF_SECS=7200
Notes:
- Backoff starts at 5 seconds and doubles on each failure
- Capped at this maximum value
- After 24 hours of failures, relay is marked "dead" and retried daily
- Lower values mean more reconnection attempts
NGIT_SYNC_STARTUP_DELAY_SECS
Description: Delay in seconds before running startup catchup
Type: Integer (seconds)
Default: 30
Required: No
Examples:
# Default: 30 second delay
NGIT_SYNC_STARTUP_DELAY_SECS=30
# Quick startup (testing)
NGIT_SYNC_STARTUP_DELAY_SECS=5
# Production: longer warm-up
NGIT_SYNC_STARTUP_DELAY_SECS=60
Notes:
- Allows connections to stabilize before catchup
- Reduces load on remote relays at startup
- Set to 0 for immediate catchup (not recommended)
NGIT_SYNC_RECONNECT_DELAY_SECS
Description: Delay in seconds before running catchup after reconnection
Type: Integer (seconds)
Default: 10
Required: No
Examples:
# Default: 10 second delay
NGIT_SYNC_RECONNECT_DELAY_SECS=10
# Quick reconnect catchup
NGIT_SYNC_RECONNECT_DELAY_SECS=5
# Conservative
NGIT_SYNC_RECONNECT_DELAY_SECS=30
Notes:
- Prevents rate limiting from remote relays
- Applied after each successful reconnection
- Only catches up on recent events (see lookback days)
NGIT_SYNC_RECONNECT_LOOKBACK_DAYS
Description: Number of days to look back for reconnect catchup
Type: Integer (days)
Default: 3
Required: No
Examples:
# Default: 3 days lookback
NGIT_SYNC_RECONNECT_LOOKBACK_DAYS=3
# Short lookback (frequent reconnects expected)
NGIT_SYNC_RECONNECT_LOOKBACK_DAYS=1
# Extended lookback
NGIT_SYNC_RECONNECT_LOOKBACK_DAYS=7
Notes:
- Limits catchup queries to recent events only
- Reduces load compared to full historical sync
- Balance between completeness and performance
- Longer lookback useful for less reliable connections
Rejected Events Index Configuration
These options configure the two-tier rejected events index that prevents wasteful re-fetching during sync and enables race condition resolution.
NGIT_REJECTED_HOT_CACHE_DURATION_SECS
Description: Duration in seconds to retain full events in hot cache for immediate re-processing
Type: Integer (seconds)
Default: 120 (2 minutes)
Required: No
Examples:
# Default: 2 minute hot cache
NGIT_REJECTED_HOT_CACHE_DURATION_SECS=120
# Shorter window (1 minute)
NGIT_REJECTED_HOT_CACHE_DURATION_SECS=60
# Longer window (5 minutes)
NGIT_REJECTED_HOT_CACHE_DURATION_SECS=300
Notes:
- Rejected events are inserted into the hot cache and cold index at the same time
- The hot cache stores full event objects for immediate re-processing when dependencies arrive
- After the full event expires, dependency-resolvable announcement and state IDs remain in the cold index and can be fetched directly from maintainer-chain relays during reciprocal invitation bootstrap
- Shorter durations reduce memory usage but may add an exact-ID relay round trip and its associated latency
- Longer durations increase memory usage and make immediate, network-free recovery more likely
- Memory impact: ~200 KB typical, ~20 MB worst case
NGIT_REJECTED_COLD_INDEX_EXPIRY_SECS
Description: Duration in seconds to retain event metadata in cold index for negentropy sync exclusion
Type: Integer (seconds)
Default: 604800 (7 days)
Required: No
Examples:
# Default: 7 day cold index
NGIT_REJECTED_COLD_INDEX_EXPIRY_SECS=604800
# Shorter retention (3 days)
NGIT_REJECTED_COLD_INDEX_EXPIRY_SECS=259200
# Longer retention (14 days)
NGIT_REJECTED_COLD_INDEX_EXPIRY_SECS=1209600
Notes:
- Cold index stores only metadata (event ID, pubkey, identifier, rejection reason)
- Prevents re-downloading rejected events during negentropy sync
- Retains the exact IDs required to recover dependency-resolvable events after their hot-cache copies expire
- Successful recovery removes an ID; failed requests leave it indexed for a throttled retry
- Entries automatically cleaned up daily
- The retention window must be long enough to cover expected invitation and maintainer-chain bootstrap delays
- Longer durations preserve more recovery opportunities and prevent more wasteful broad re-fetching, but use slightly more memory
- Memory impact: ~1 MB typical
Holding DB and Deletion-Request Cleanup Configuration
These options control retention for deleted events archived in the holding database and the shared cleanup cadence for holding and deletion-request retention records.
NGIT_HOLDING_RETENTION_SECS
Description: How long deleted events are retained in the holding DB before expiration cleanup removes them
Type: Integer (seconds)
Default: 7776000 (90 days)
Required: No
Examples:
# Default: 90 days
NGIT_HOLDING_RETENTION_SECS=7776000
# Short retention (30 days)
NGIT_HOLDING_RETENTION_SECS=2592000
Notes:
- Must be greater than 0
- Used by both startup catch-up cleanup and periodic cleanup passes
NGIT_HOLDING_CLEANUP_INTERVAL_SECS
Description: Interval between periodic holding DB expiration cleanup passes and deletion-request retention cleanup passes
Type: Integer (seconds)
Default: 86400 (24 hours)
Required: No
Examples:
# Default: daily cleanup
NGIT_HOLDING_CLEANUP_INTERVAL_SECS=86400
# Every 6 hours
NGIT_HOLDING_CLEANUP_INTERVAL_SECS=21600
Notes:
- Must be greater than 0
- Smaller values clean up expired records sooner at the cost of more background work
- This interval determines how soon expired records are observed and physically removed. A deletion request may remain queryable from Main until the next cleanup pass, but admission-gate eligibility still ends at its timestamp-derived deadline; this cleanup latency does not reset or extend the lifecycle.
GRASP-05 Archive Configuration
These options enable archive/mirror/backup mode per the GRASP-05 specification.
NGIT_ARCHIVE_ALL
Description: Accept all repository announcements regardless of whether they list this instance
Type: Boolean
Default: false
Required: No
Examples:
# Enable archive-all mode (⚠️ WARNING: Storage risk)
NGIT_ARCHIVE_ALL=true
# Disable (default - GRASP-01 strict mode)
NGIT_ARCHIVE_ALL=false
Security Warning: When enabled, any repository can be mirrored to this relay, potentially causing storage and bandwidth exhaustion. Only enable if you have unlimited resources and trust the relay network.
Notes:
- Archived repositories are read-only (pushes rejected)
- Full sync enabled (both git data and Nostr events)
- Takes precedence over whitelist (accepts everything)
NGIT_ARCHIVE_WHITELIST
Description: Comma-separated list of repositories/pubkeys/identifiers to archive
Type: String (comma-separated)
Default: (empty)
Required: No
Formats:
<npub>- Archive all repos from this pubkey<npub>/<identifier>- Archive specific repo from specific pubkey<identifier>- Archive repos with this identifier from any pubkey
Examples:
# Archive all repos from Alice
NGIT_ARCHIVE_WHITELIST=npub1alice23
# Archive specific repos
NGIT_ARCHIVE_WHITELIST=npub1alice23/linux,npub1bob23/bitcoin-core
# Archive by identifier (any pubkey)
NGIT_ARCHIVE_WHITELIST=bitcoin-core,linux,rust
# Mixed formats
NGIT_ARCHIVE_WHITELIST=npub1alice23...,npub1bob23.../linux,bitcoin-core
Validation:
- Npub entries are validated at startup (invalid npub = server fails to start)
- Identifier entries accept any string
- Whitespace is trimmed
- Empty entries are ignored
Security Notes:
- Identifier-only format (
bitcoin-core) matches ANY pubkey - Use
npub/identifierformat for high-value archives - Whitelist is static (restart required to change)
- Future: Dynamic management via API
NGIT_ARCHIVE_GRASP_SERVICES
Description: Comma-separated list of GRASP server domains to archive
Type: String (comma-separated domain names)
Default: (empty)
Required: No
Format:
<domain>- Archive all repositories from this GRASP server domain- Must be bare domains only (e.g.,
git.example.com, NOTwss://git.example.com) - Matching extracts domains from announcement clone URLs and compares them exactly (case-sensitive)
Examples:
# Archive all repos from a single GRASP server
NGIT_ARCHIVE_GRASP_SERVICES=git.example.com
# Archive repos from multiple GRASP servers
NGIT_ARCHIVE_GRASP_SERVICES=git.example.com,git.nostr.dev,relay.gitnostr.com
# Archive from localhost (testing)
NGIT_ARCHIVE_GRASP_SERVICES=localhost:7334
Validation:
- Domain entries must be bare domains without scheme prefixes (ws://, wss://, https://, etc.)
- Whitespace is trimmed
- Empty entries are ignored
- Mutually exclusive with
NGIT_ARCHIVE_ALLandNGIT_ARCHIVE_WHITELIST
Security Notes:
- Archives ALL repositories from the specified GRASP server domains
- Use with caution - ensure you trust the GRASP servers you're archiving from
- Storage requirements depend on the size of repositories on the archived servers
- Automatically sets
NGIT_ARCHIVE_READ_ONLY=trueby default
Error Conditions:
# ERROR: Cannot use with NGIT_ARCHIVE_ALL
NGIT_ARCHIVE_GRASP_SERVICES=git.example.com
NGIT_ARCHIVE_ALL=true
# → Server fails to start: "NGIT_ARCHIVE_GRASP_SERVICES cannot be used with
# NGIT_ARCHIVE_ALL=true. These options are mutually exclusive."
# ERROR: Cannot use with NGIT_ARCHIVE_WHITELIST
NGIT_ARCHIVE_GRASP_SERVICES=git.example.com
NGIT_ARCHIVE_WHITELIST=npub1alice...
# → Server fails to start: "NGIT_ARCHIVE_GRASP_SERVICES cannot be used with
# NGIT_ARCHIVE_WHITELIST. These options are mutually exclusive."
Use Cases:
# Backup/mirror a specific GRASP server
NGIT_ARCHIVE_GRASP_SERVICES=git.example.com
NGIT_ARCHIVE_READ_ONLY=true # Default
# Archive multiple trusted GRASP servers
NGIT_ARCHIVE_GRASP_SERVICES=git.nostr.dev,relay.gitnostr.com
NGIT_ARCHIVE_READ_ONLY
Description: Configure relay as read-only sync of archived repositories
Type: Boolean
Default: true if NGIT_ARCHIVE_ALL, NGIT_ARCHIVE_WHITELIST, or NGIT_ARCHIVE_GRASP_SERVICES is set, false otherwise
Required: No
Examples:
# Explicitly enable (requires archive mode)
NGIT_ARCHIVE_READ_ONLY=true
# Explicitly disable (writable archive repos)
NGIT_ARCHIVE_READ_ONLY=false
# Automatic (default behavior)
# - If NGIT_ARCHIVE_ALL, NGIT_ARCHIVE_WHITELIST, or NGIT_ARCHIVE_GRASP_SERVICES is set → true
# - Otherwise → false
# NGIT_ARCHIVE_READ_ONLY=
Behavior:
- When
true:- NIP-11 document includes
GRASP-05insupported_grasps - NIP-11
curationfield describes the archive scope - Repository announcements not listing this service are accepted per whitelist/archive-all
- NIP-11 document includes
- When
false:- Archive mode disabled (standard GRASP-01 operation)
- When unset (default):
- Automatically
trueif archive mode configured - Automatically
falseotherwise
- Automatically
Error Conditions:
# ERROR: Cannot set read-only without archive config
NGIT_ARCHIVE_READ_ONLY=true
NGIT_ARCHIVE_ALL=false
NGIT_ARCHIVE_WHITELIST=
NGIT_ARCHIVE_GRASP_SERVICES=
# → Server fails to start: "NGIT_ARCHIVE_READ_ONLY=true requires either
# NGIT_ARCHIVE_ALL=true, NGIT_ARCHIVE_WHITELIST, or NGIT_ARCHIVE_GRASP_SERVICES to be set"
# ERROR: Cannot use repository whitelist with archive read-only
NGIT_ARCHIVE_READ_ONLY=true
NGIT_ARCHIVE_WHITELIST=npub1alice...
NGIT_REPOSITORY_WHITELIST=npub1bob...
# → Server fails to start: "NGIT_REPOSITORY_WHITELIST cannot be used with
# NGIT_ARCHIVE_READ_ONLY=true"
NIP-11 Impact:
When NGIT_ARCHIVE_READ_ONLY=true:
supported_grasps: includes"GRASP-05"curation: Set to one of:"Read-only sync of all repositories found on network"(ifNGIT_ARCHIVE_ALL=true)"Read-only sync of whitelisted repositories and maintainers"(ifNGIT_ARCHIVE_WHITELISTset)"Read-only sync of repositories from specified GRASP servers"(ifNGIT_ARCHIVE_GRASP_SERVICESset)
Use Cases:
# Public archive of entire ecosystem
NGIT_ARCHIVE_ALL=true
NGIT_ARCHIVE_READ_ONLY=true # Default
# Selective backup of critical projects
NGIT_ARCHIVE_WHITELIST=npub1torvalds.../linux,npub1satoshi.../bitcoin
NGIT_ARCHIVE_READ_ONLY=true # Default
# Writable mirror (advanced, not typical)
NGIT_ARCHIVE_WHITELIST=npub1alice...
NGIT_ARCHIVE_READ_ONLY=false
# Archive specific GRASP servers
NGIT_ARCHIVE_GRASP_SERVICES=git.example.com,git.nostr.dev
NGIT_ARCHIVE_READ_ONLY=true # Default
GRASP-06 Contributor PR Submission
These options control the optional /prs/<npub>/<identifier>.git contributor pull-request submission endpoint per the GRASP-06 specification. The route is below NGIT_BASE_PATH when a non-root mount is configured.
NGIT_GRASP06_ENABLE
Description: Enable the GRASP-06 contributor PR submission endpoint at /prs/<npub>/<identifier>.git
Type: Boolean
Default: false
Required: No
Examples:
# Enable GRASP-06 (opt-in)
NGIT_GRASP06_ENABLE=true
# Disable (default)
NGIT_GRASP06_ENABLE=false
Behavior:
- When
true:/prs/<npub>/<identifier>.gitaccepts unauthenticated pushes ofrefs/nostr/<event-id>- NIP-11
supported_graspsincludes"GRASP-06" - PR / PR-Update events naming this relay's
/prs/endpoint in theirclonetag are accepted to purgatory even without a matching accepted repository announcement - Standard
<npub>/<identifier>.gitendpoint behaviour is unchanged
- When
false(default):/prs/*returns HTTP 404- Event-acceptance is unchanged (standard GRASP-01 rules apply)
Security Model:
The /prs/ endpoint is intentionally unauthenticated — there is no NIP-98 auth, no allowlist, no quota, and no proof-of-work in v1. Validity is enforced by:
- The signed PR / PR-Update Nostr event the pushed
refs/nostr/<event-id>references - Per-submitter / per-identifier scoping enforced when the event arrives
- Inline ref-name validation (only
refs/nostr/<64-hex-event-id>accepted) - Periodic cleanup of orphan repositories with zero refs
Operators should review the design tradeoffs in docs/explanation/grasp-06-contributor-pr-submission.md before enabling.
GRASP-08 Private Service
NGIT_PRIVATE_MODE
Description: Enable authenticated, read-restricted GRASP-08 service mode
Type: Boolean
Default: false
Required: No
When enabled, the relay requires successful NIP-42 authentication by a
whitelisted member before accepting or serving any Nostr events. Standard Git
repository root, info/refs, git-upload-pack, and git-receive-pack
requests require the GRASP-08 repository-scoped NIP-98 credential. The
relay-owner identity events (kind 0/10002) are still seeded and served
locally but are never published to NGIT_USER_INDEX_RELAYS, so a private
relay does not advertise its existence.
NGIT_PRIVATE_MODE=true
Private mode cannot be combined with NGIT_GRASP06_ENABLE=true, because the
GRASP-06 contributor endpoint is intentionally unauthenticated.
NGIT_PRIVATE_MEMBERS
Description: Permanently configured members of the service-wide private-service whitelist
Type: Comma-separated npubs
Default: Empty
Required: Yes when NGIT_PRIVATE_MODE=true
NGIT_PRIVATE_MEMBERS=npub1alice...,npub1bob...
The effective GRASP-08 whitelist is the union of these configured members and the NIP-11 owner pubkeys of relays referenced by accepted repository announcements. Relay owners are learned by the NIP-11 request already made for sync limits; this does not open another connection or subscription. A relay listed only by an unpromoted purgatory announcement cannot grant access.
NGIT_PRIVATE_PUBLIC_ORIGIN
Optional canonical external HTTP origin for private Git authentication, such
as https://private.example. Configure it when TLS terminates at a reverse
proxy or when a non-loopback deployment intentionally uses plain HTTP. The
value must not include a path, query, or fragment. If omitted, the service
retains the legacy inference of HTTP for loopback domains and HTTPS otherwise.
Whitespace and empty comma-separated elements are ignored. Every non-empty entry must be a valid npub. Invalid entries fail startup rather than being silently skipped because this setting controls access, not repository curation. Changing the environment setting requires a service restart.
Repository Whitelist
NGIT_REPOSITORY_WHITELIST
Description: Whitelist specific repositories/pubkeys/identifiers for GRASP-01 acceptance
Type: Comma-separated list
Default: Empty (all repos listing our service are accepted)
Required: No
Format: Same as NGIT_ARCHIVE_WHITELIST:
npub1...- Accept all repos from this pubkey (if they list our service)npub1.../identifier- Accept specific repo (if it lists our service)identifier- Accept repos with this identifier (if they list our service)
Difference from Archive Whitelist:
- Repository whitelist: Announcements MUST list our service AND match whitelist
- Archive whitelist: Announcements don't need to list our service, just match whitelist
Examples:
# Accept only repos from specific pubkey (that list our service)
NGIT_REPOSITORY_WHITELIST=npub1alice23
# Accept specific repos only
NGIT_REPOSITORY_WHITELIST=npub1alice23/linux,npub1bob23/bitcoin-core
# Accept repos with specific identifiers
NGIT_REPOSITORY_WHITELIST=bitcoin-core,linux,rust
# Combined whitelist
NGIT_REPOSITORY_WHITELIST=npub1alice23...,npub1bob23.../linux,bitcoin-core
Behavior:
- When set:
- Announcements must list our service in both
cloneandrelaystags (GRASP-01 requirement) - Announcements must match the whitelist (pubkey, repo, or identifier)
- NIP-11
curationfield set to:"Accepts only whitelisted repositories and maintainers that list this service"
- Announcements must list our service in both
- When empty (default):
- All announcements listing our service are accepted (standard GRASP-01 behavior)
Error Conditions:
# ERROR: Cannot use with archive read-only mode
NGIT_ARCHIVE_READ_ONLY=true
NGIT_ARCHIVE_WHITELIST=npub1archive...
NGIT_REPOSITORY_WHITELIST=npub1bob...
# → Server fails to start: "NGIT_REPOSITORY_WHITELIST cannot be used with
# NGIT_ARCHIVE_READ_ONLY=true. Either set NGIT_ARCHIVE_READ_ONLY=false
# or use NGIT_ARCHIVE_WHITELIST instead"
NIP-11 Impact:
When NGIT_REPOSITORY_WHITELIST is set:
curation:"Accepts only whitelisted repositories and maintainers that list this service"supported_grasps: Does not includeGRASP-05(still GRASP-01 compliant)
Use Cases:
# Curated relay for specific projects (GRASP-01 mode)
NGIT_REPOSITORY_WHITELIST=bitcoin-core,linux,rust
# Personal relay for self and trusted collaborators
NGIT_REPOSITORY_WHITELIST=npub1me...,npub1alice...,npub1bob...
# Project-specific relay (e.g., Rust ecosystem)
NGIT_REPOSITORY_WHITELIST=rust,cargo,rustc,tokio,serde
# Hybrid: specific projects AND specific maintainer's repos
NGIT_REPOSITORY_WHITELIST=bitcoin-core,npub1alice...
Comparison Table:
| Configuration | Lists Service? | Matches Whitelist? | Result |
|---|---|---|---|
| No whitelist | Yes | N/A | ✅ Accept (GRASP-01) |
| No whitelist | No | N/A | ❌ Reject |
| Repository whitelist | Yes | Yes | ✅ Accept (GRASP-01) |
| Repository whitelist | Yes | No | ❌ Reject (not whitelisted) |
| Repository whitelist | No | Yes | ❌ Reject (doesn't list service) |
| Archive whitelist (read-only=true) | No | Yes | ✅ Accept (GRASP-05) |
| Archive whitelist (read-only=false) | Yes | N/A | ✅ Accept (GRASP-01) |
| Archive whitelist (read-only=false) | No | Yes | ✅ Accept (GRASP-05) |
Repository Blacklist
NGIT_REPOSITORY_BLACKLIST
Description: Blacklist specific repositories/pubkeys/identifiers to reject
Type: Comma-separated list
Default: Empty (no repositories are blacklisted)
Required: No
Format: Same as whitelist formats:
npub1...- Block all repos from this pubkeynpub1.../identifier- Block specific repoidentifier- Block repos with this identifier (any pubkey)
Precedence: Blacklist takes precedence over ALL whitelists:
- Blacklisted repos are rejected even if they match archive or repository whitelists
- Blacklisted repos are rejected even if they list our service
- Blacklist is checked first before any other validation
Examples:
# Block all repos from specific pubkey
NGIT_REPOSITORY_BLACKLIST=npub1spam...
# Block specific repo
NGIT_REPOSITORY_BLACKLIST=npub1alice.../malware-repo
# Block repos with specific identifiers
NGIT_REPOSITORY_BLACKLIST=malware,spam,phishing
# Combined blacklist
NGIT_REPOSITORY_BLACKLIST=npub1spam...,npub1alice.../bad-repo,malware
Rejection Reasons:
The blacklist provides specific rejection reasons based on the match type:
- Npub format:
"Repository owner <npub> is blacklisted" - Npub/identifier format:
"Repository <npub>/<identifier> is blacklisted" - Identifier format:
"Repository identifier <identifier> is blacklisted"
These reasons help operators understand why a repository was rejected without needing to flag it in curation metadata.
Behavior:
Blacklist is checked before all other validation:
- Check blacklist → Reject if matched
- Check if lists service → Accept if matches repository whitelist (if enabled)
- Check archive config → Accept if matches archive whitelist (if enabled)
- Reject otherwise
Use Cases:
# Block spam/malware repos
NGIT_REPOSITORY_BLACKLIST=malware,spam,phishing
# Block abusive users
NGIT_REPOSITORY_BLACKLIST=npub1spammer...,npub1abuser...
# Block specific problematic repos
NGIT_REPOSITORY_BLACKLIST=npub1alice.../copyright-violation,npub1bob.../illegal-content
# Temporary block for investigation
NGIT_REPOSITORY_BLACKLIST=npub1suspicious.../repo-under-review
Comparison with Whitelists:
| Configuration | Blacklisted? | Matches Whitelist? | Lists Service? | Result |
|---|---|---|---|---|
| Blacklist only | Yes | N/A | N/A | ❌ Reject (blacklisted) |
| Blacklist only | No | N/A | Yes | ✅ Accept (GRASP-01) |
| Blacklist + Repository whitelist | Yes | Yes | Yes | ❌ Reject (blacklist wins) |
| Blacklist + Archive whitelist | Yes | Yes | No | ❌ Reject (blacklist wins) |
| Blacklist + Both whitelists | Yes | Yes | Yes | ❌ Reject (blacklist wins) |
| Blacklist only | No | N/A | No | ❌ Reject (no whitelist match) |
NIP-11 Impact:
Blacklist does not affect NIP-11 metadata:
- No
curationfield changes (blacklist is operational, not curation policy) - Blacklist is transparent to clients (rejected with specific reason)
- Operators can use blacklist without advertising curation
NGIT_BLACKLIST_AUTO_RESTORE
Description: On startup, restore repositories previously deleted by blacklist parity when they are no longer blacklisted and still within holding retention
Type: Boolean
Default: false
Required: No
CLI: --blacklist-auto-restore
Behavior:
- When
false(default): startup does not run blacklist-source auto-restore; unblacklisted repositories in holding are restored only by existing recovery triggers (for example re-announcement). - When
true: startup also scans holding metadata withholding-source=blacklist, deduplicates owner+identifier scopes, and attempts restore for scopes that:- are still within
NGIT_HOLDING_RETENTION_SECS, and - no longer match
NGIT_REPOSITORY_BLACKLIST.
- are still within
- Scopes that are still blacklisted are skipped (not restored).
Related startup behavior (always on when repository whitelist is enabled):
- Startup scans holding metadata with
holding-source=whitelistand attempts restore for scopes that are within retention, now matchNGIT_REPOSITORY_WHITELIST, and are not blacklisted. - Scopes still not matching
NGIT_REPOSITORY_WHITELISTare skipped.
Startup ordering:
- Blacklist parity delete
- Blacklist auto-restore (if enabled)
- Whitelist parity delete
- Whitelist restore (for scopes now matching
NGIT_REPOSITORY_WHITELIST)
This ordering keeps startup reconciliation deterministic when multiple policies interact.
Examples:
# Enable startup auto-restore for unblacklisted repos
NGIT_BLACKLIST_AUTO_RESTORE=true
# Default behavior
NGIT_BLACKLIST_AUTO_RESTORE=false
Event Blacklist
NGIT_EVENT_BLACKLIST
Description: Blacklist events from specific authors (npubs)
Type: Comma-separated list of npubs
Default: Empty (no events are blacklisted by author)
Required: No
Format:
npub1...- Block all events from this author
Precedence: Event blacklist takes precedence over ALL other validation:
- Blacklisted events are rejected before any other policy checks
- Applies to all event types (announcements, state events, PRs, etc.)
- Events never reach purgatory (rejected immediately)
- Overrides repository blacklist, whitelists, and all other policies
Examples:
# Block all events from specific author
NGIT_EVENT_BLACKLIST=npub1spam...
# Block events from multiple authors
NGIT_EVENT_BLACKLIST=npub1spam...,npub1abuser...,npub1troll...
Rejection Reason:
The event blacklist provides a specific rejection reason:
- Format:
"Event author <npub> is blacklisted"
This reason helps operators understand why an event was rejected without needing to flag it in metadata.
Behavior:
Event blacklist is checked first before all other validation:
- Check event blacklist → Reject if author is blacklisted
- Check repository blacklist (for announcements) → Reject if matched
- Check event-type specific policies → Accept/Reject based on policy
- Process event normally
Use Cases:
# Block spam/abusive users
NGIT_EVENT_BLACKLIST=npub1spammer...,npub1abuser...
# Block malicious actors
NGIT_EVENT_BLACKLIST=npub1malware...,npub1phisher...
# Temporary block for investigation
NGIT_EVENT_BLACKLIST=npub1suspicious...
Comparison with Repository Blacklist:
| Configuration | Scope | Checked When | Applies To |
|---|---|---|---|
| Event Blacklist | Author-based | First (before all policies) | All events from author |
| Repository Blacklist | Repo-based | Second (announcements only) | Specific repositories |
Event Blacklist vs Repository Blacklist:
# Scenario: npub1alice is event-blacklisted
NGIT_EVENT_BLACKLIST=npub1alice...
# Result:
# - ALL events from npub1alice are rejected (announcements, PRs, etc.)
# - Events never reach relay or purgatory
# - Rejection: "Event author npub1alice... is blacklisted"
# Scenario: npub1alice/repo is repository-blacklisted
NGIT_REPOSITORY_BLACKLIST=npub1alice.../malware
# Result:
# - Only announcements for npub1alice.../malware are rejected
# - Other events from npub1alice are still processed normally
# - PRs/state events for different repos from npub1alice are accepted
NIP-11 Impact:
Event blacklist does not affect NIP-11 metadata:
- No
curationfield changes (blacklist is operational, not policy) - Blacklist is transparent to clients (rejected with specific reason)
- Operators can use blacklist without advertising moderation
Deletion Requests (NIP-09 and NIP-62)
NGIT_DELETION_REQUEST_DISRESPECTOR
Description: Ignore NIP-09 deletion requests and NIP-62 request-to-vanish events and act as an archival server
Type: Boolean
Default: false (deletion requests are honoured)
Required: No
CLI: --deletion-request-disrespector
Behavior:
- When
false(default):- NIP-09 (kind 5) deletion requests and NIP-62 request-to-vanish events are honoured: their targets are hard-deleted from the relay, matching purgatory entries are evicted, and a persistent tombstone is recorded so re-submission of the deleted event stays rejected across restarts.
- NIP-11
supported_nipsincludes9(deletion) and62(request to vanish).
- When
true:- Incoming NIP-09 deletion requests and NIP-62 request-to-vanish events are still stored (the client receives an OK), but they are not acted upon. Their targets remain fully accessible. This makes the relay an archival server, preserving content and preventing "left-pad" scenarios.
- NIP-11
supported_nipsdoes not include9or62, so clients can discover that this relay does not honour deletions.
IMPORTANT: This setting ONLY affects NIP-09 and NIP-62 user-initiated requests. It does NOT prevent blacklist-triggered deletions, which are an operator moderation mechanism (spam/malware/abuse) that archival relays still need.
Use Cases:
- Community archival relays
- Research / historical preservation
- Backup / mirror relays
Examples:
# Archival relay: preserve targets of NIP-09 and NIP-62 requests
NGIT_DELETION_REQUEST_DISRESPECTOR=true
# Standard relay: honour deletion requests (default)
NGIT_DELETION_REQUEST_DISRESPECTOR=false
See docs/explanation/repository-lifecycle.md
for the full lifecycle design rationale.
Deletion-Request Retention (NIP-09 and NIP-62)
These options govern the bounded lifecycle of accepted NIP-09 deletion requests and NIP-62 request-to-vanish events. All values are integer seconds. The used-request defaults express months as fixed 30-day periods, not calendar months.
The relay derives unused deadlines from relay-observed first_seen_at, never from the client-controlled event created_at. It derives used deadlines from last_used_at. Each ...ADDITIONAL... option begins only after its corresponding served period ends; it is not a total retention duration.
Seconds are used for configuration consistency and short automated tests. In production, configure every period to at least one day and keep each total lifecycle comfortably longer than the worst-case deletion processing time, including repository archival and cascade work. Sub-day values are intended only for tests.
NGIT_DELETION_REQUEST_RETENTION_UNUSED_SERVED_SECS
- Description: How long an unused deletion or vanish request remains served from relay-observed
first_seen_at - Type: Positive integer (seconds)
- Default:
2592000(30 days) - Required: No
- CLI:
--deletion-request-retention-unused-served-secs
# Default: serve an unused request for 30 days from first_seen_at
NGIT_DELETION_REQUEST_RETENTION_UNUSED_SERVED_SECS=2592000
After this served period, an unused request enters its additional unserved/gating period.
NGIT_DELETION_REQUEST_RETENTION_UNUSED_UNSERVED_GATING_ADDITIONAL_SECS
- Description: Additional time after unused serving ends that an unused deletion or vanish request remains unserved but eligible to gate admission
- Type: Positive integer (seconds)
- Default:
15552000(180 days) - Required: No
- CLI:
--deletion-request-retention-unused-unserved-gating-additional-secs
# Default: retain an unused request as an unserved gate for a further 180 days
NGIT_DELETION_REQUEST_RETENTION_UNUSED_UNSERVED_GATING_ADDITIONAL_SECS=15552000
The unused request expires after unused served + this additional period from first_seen_at. Retained disrespector and non-targeting NIP-62 records do not enforce a local admission gate. When an archival relay starts, it reclassifies an expired unused targeting request as used if its target is present before cleanup runs; this preserves the used request indefinitely.
NGIT_DELETION_REQUEST_RETENTION_USED_SERVED_AFTER_LAST_USED_SECS
- Description: In normal mode, how long a used deletion or vanish request remains served after
last_used_at - Type: Positive integer (seconds)
- Default:
23328000(270 days; 9 fixed 30-day months) - Required: No
- CLI:
--deletion-request-retention-used-served-after-last-used-secs
# Default: serve a used request for 270 days after last_used_at
NGIT_DELETION_REQUEST_RETENTION_USED_SERVED_AFTER_LAST_USED_SECS=23328000
Every successful use resets last_used_at and restarts this served period. When NGIT_DELETION_REQUEST_DISRESPECTOR=true, used requests remain served indefinitely, so this normal-mode duration does not expire them.
NGIT_DELETION_REQUEST_RETENTION_USED_UNSERVED_GATING_ADDITIONAL_SECS
- Description: Additional normal-mode time after used serving ends that a used deletion or vanish request remains unserved but continues gating admission
- Type: Positive integer (seconds)
- Default:
7776000(90 days; 3 fixed 30-day months) - Required: No
- CLI:
--deletion-request-retention-used-unserved-gating-additional-secs
# Default: continue gating for a further 90 days after used serving ends
NGIT_DELETION_REQUEST_RETENTION_USED_UNSERVED_GATING_ADDITIONAL_SECS=7776000
The used request expires after used served + this additional period from last_used_at. When NGIT_DELETION_REQUEST_DISRESPECTOR=true, used requests remain served indefinitely, so this normal-mode additional period does not expire their local record.
Validation: Each deletion-request retention duration must be greater than zero. Each served and additional-gating pair must also fit in an unsigned 64-bit second duration when combined.
Cleanup cadence: NGIT_HOLDING_CLEANUP_INTERVAL_SECS schedules both holding DB and deletion-request retention cleanup passes. Cleanup evaluates deadlines derived from first_seen_at and last_used_at; changing its cadence never changes those deadlines.
Rate Limiting & DoS Protection
NGIT_MAX_CONNECTIONS
Description: Maximum total connections to the relay. When unset, connections are unlimited, deferring to OS fd limits and infrastructure-level controls.
Type: Integer
Default: unlimited
Required: No
Examples:
# Cap connections for a resource-constrained deployment
NGIT_MAX_CONNECTIONS=4096
# Higher limit for large public relay
NGIT_MAX_CONNECTIONS=8000
# Lower limit for private relay
NGIT_MAX_CONNECTIONS=100
Notes:
- When unset, the relay imposes no connection limit (
Semaphore::MAX_PERMITS); OS fd limits and infrastructure controls apply - Set this only if you need an explicit cap; otherwise leave unset
- Works in conjunction with per-connection limits (500 subscriptions, 60 events/min)
- When limit is reached, new connections are rejected
- Existing connections continue to work normally
Embedded relay resource limits
rust-nostr 0.45 introduced or tightened several local-relay defaults. ngit-grasp sets every effective value explicitly in code, but exposes only limits that a peer can discover and use to adapt sync, plus the Git-specific event-size policy.
| Environment variable | Default | Meaning |
|---|---|---|
NGIT_RELAY_MAX_SUBSCRIPTIONS |
500 |
Active REQ subscriptions per WebSocket connection |
NGIT_RELAY_FILTER_LIMIT |
500 |
Explicit and omitted-limit results per filter |
NGIT_RELAY_MAX_EVENT_SIZE_BYTES |
196608 |
Serialized event bytes (192 KiB) |
Every value must be greater than zero. The event limit must not exceed the fixed 5 MiB WebSocket message bound. Its default is three times rust-nostr's 64 KiB default because production history contains a valid NIP-34 patch event of about 149 KiB; the bound retains approximately 29% headroom.
The NIP-11 document advertises NGIT_RELAY_MAX_SUBSCRIPTIONS as
max_subscriptions, and NGIT_RELAY_FILTER_LIMIT as both max_limit and
default_limit. ngit-grasp peers already consume these fields for their
per-connection subscription ledger and adaptive pagination. Event size has no
equivalent standard NIP-11 field, so it remains documented configuration only.
Logging Configuration
NGIT_LOG_LEVEL
Description: Application logging level or explicit tracing filter expression
Type: String (log level or filter expression)
Default: info
Required: No
Examples:
# Simple levels
NGIT_LOG_LEVEL=error # ngit-grasp errors; dependency warnings and errors
NGIT_LOG_LEVEL=warn # ngit-grasp warnings and errors; dependency warnings and errors
NGIT_LOG_LEVEL=info # ngit-grasp info and above; dependency warnings and errors (default)
NGIT_LOG_LEVEL=debug # ngit-grasp debug and above; dependency warnings and errors
NGIT_LOG_LEVEL=trace # ngit-grasp trace and above; dependency warnings and errors
# Module-specific filtering
NGIT_LOG_LEVEL=ngit_grasp=debug,actix_web=info
# Complex filters
NGIT_LOG_LEVEL=debug,hyper=info,tokio=warn
Bare levels are expanded to warn,ngit_grasp=<level> so application
diagnostics do not also enable verbose internals from every dependency.
Expressions containing target directives or commas are passed through
unchanged, leaving dependency verbosity under explicit operator control.
Log levels (most to least verbose):
trace- Very detailed, performance impactdebug- Detailed debugging informationinfo- General information (default)warn- Warnings about potential issueserror- Errors only
CLI flag:
ngit-grasp --log-level trace
Production recommendation:
NGIT_LOG_LEVEL=info
Notes:
- Uses Rust's
tracingcrate filter syntax - Bare levels apply to ngit-grasp while dependencies remain at
warn - Supports module-level filtering (e.g.,
ngit_grasp=debug,hyper=info) tracelevel can significantly impact performance- For production,
infoorwarnis recommended
Security Configuration (Planned)
NGIT_AUTH_REQUIRED
Description: Require authentication for all operations
Type: Boolean
Default: false
Status: 🔜 Planned
Examples:
NGIT_AUTH_REQUIRED=true # Require auth
NGIT_AUTH_REQUIRED=false # Public relay
NGIT_RATE_LIMIT_ENABLED
Description: Enable rate limiting
Type: Boolean
Default: true
Status: 🔜 Planned
Examples:
NGIT_RATE_LIMIT_ENABLED=true
NGIT_RATE_LIMIT_ENABLED=false
Configuration File (.env)
For development, create a .env file in the project root:
# .env file example
NGIT_DOMAIN=localhost:7334
NGIT_RELAY_OWNER_NSEC=nsec1...
NGIT_RELAY_NAME="Development Relay"
NGIT_RELAY_DESCRIPTION="Local development instance"
NGIT_GIT_DATA_PATH=./data/git
NGIT_RELAY_DATA_PATH=./data/relay
NGIT_BIND_ADDRESS=127.0.0.1:7334
RUST_LOG=debug
Notes:
- Never commit
.envto version control - Use
.env.exampleas a template - Environment variables override
.envvalues
Validation
Configuration is validated at startup:
// Example validation errors:
Error: Invalid configuration
- NGIT_DOMAIN is required
- Invalid relay owner nsec in NGIT_RELAY_OWNER_NSEC
- NGIT_GIT_DATA_PATH is not writable
Validation checks:
- Required fields are present
- Values have correct format
- Paths are accessible and writable
- Ports are available
- Relay owner nsec is valid
Production Configuration Example
# Production .env
NGIT_DOMAIN=gitnostr.com
NGIT_RELAY_OWNER_NSEC=nsec1...
NGIT_RELAY_NAME="GitNostr Public Relay"
NGIT_RELAY_DESCRIPTION="Public GRASP relay for open source projects"
NGIT_GIT_DATA_PATH=/var/lib/ngit-grasp/git
NGIT_RELAY_DATA_PATH=/var/lib/ngit-grasp/relay
NGIT_BIND_ADDRESS=127.0.0.1:7334
NGIT_TRUSTED_PROXY_CIDRS=127.0.0.1/32
RUST_LOG=info,ngit_grasp=debug
Additional production considerations:
- Use reverse proxy (nginx, Caddy) for HTTPS
- Keep the backend private and set
NGIT_TRUSTED_PROXY_CIDRSto the actual proxy source networks if client-IP accounting is required - Set up log rotation
- Configure monitoring
- Implement backup strategy
- Use dedicated user account
- Set file permissions properly
Development Configuration Example
# Development .env
NGIT_DOMAIN=localhost:7334
NGIT_RELAY_OWNER_NSEC=nsec1...
NGIT_RELAY_NAME="Dev Relay"
NGIT_RELAY_DESCRIPTION="Local development"
NGIT_GIT_DATA_PATH=./data/git
NGIT_RELAY_DATA_PATH=./data/relay
NGIT_BIND_ADDRESS=127.0.0.1:7334
RUST_LOG=debug
Testing Configuration Example
# Testing .env
NGIT_DOMAIN=localhost:9999
NGIT_RELAY_OWNER_NSEC=nsec1...
NGIT_RELAY_NAME="Test Relay"
NGIT_RELAY_DESCRIPTION="Automated testing"
NGIT_GIT_DATA_PATH=/tmp/ngit-test/git
NGIT_RELAY_DATA_PATH=/tmp/ngit-test/relay
NGIT_BIND_ADDRESS=127.0.0.1:9999
RUST_LOG=debug
Testing notes:
- Use temporary directories
- Use non-standard ports
- Clean up after tests
- Isolate from development data
Configuration Priority
When multiple configuration sources exist:
- Command-line arguments (highest priority, planned)
- Environment variables
.envfile- Default values (lowest priority)
Example:
# .env file
NGIT_BIND_ADDRESS=127.0.0.1:7334
# Environment variable (overrides .env)
NGIT_BIND_ADDRESS=0.0.0.0:3000 cargo run
# Result: binds to 0.0.0.0:3000
Related Documentation
- Deployment How-To - Production deployment
- Getting Started Tutorial - Initial setup
- Architecture Overview - System design
Part of the ngit-grasp reference documentation