ptr.pet
at protocol indexer with flexible filtering, xrpc queries, and a cursor-backed event stream, built on fjall
at-protocol
atproto
indexer
rust
fjall
[docs] update agents.md
+17
-18
AGENTS.md
+17
-18
AGENTS.md
···
22
23
## Project overview
24
25
-
Hydrant is an AT Protocol indexer built on the `fjall` LSM-tree engine. It supports both full-network indexing and efficient targeted indexing (filtered by DID), while maintaining full Firehose compatibility.
26
27
Key design goals:
28
- Ingestion via the `fjall` storage engine.
29
- Content-Addressable Storage (CAS) for IPLD blocks.
30
- Reliable backfill mechanism with buffered live-event replay.
31
- Efficient binary storage using MessagePack (`rmp-serde`).
32
-
- Native integration with the `jacquard` suite of ATProto crates.
33
34
## System architecture
35
36
-
Hydrant consists of several concurrent components:
37
-
- **Ingestor**: Connects to an upstream Firehose (Relay) and filters events. It manages the transition between discovery and synchronization.
38
-
- **Crawler**: Periodically enumerates the network via `com.atproto.sync.listRepos` to discover new repositories when in full-network mode.
39
-
- **Backfill worker**: A dedicated worker that fetches full repository CAR files from PDS instances when a new repo is detected.
40
-
- **API server**: An Axum-based XRPC server implementing repository read methods (`getRecord`, `listRecords`) and system stats. It also provides a TAP-compatible JSON stream API via WebSockets.
41
-
- **Persistence worker**: Manages periodic background flushes of the LSM-tree and cursor state.
42
43
### Lazy event inflation
44
To minimize latency in `apply_commit` and the backfill worker, events are stored in a compact `StoredEvent` format. The expansion into full TAP-compatible JSON (including fetching record content from the CAS and DAG-CBOR parsing) is performed lazily within the WebSocket stream handler.
45
46
## General conventions
···
52
- Prefer compile-time guarantees over runtime checks where possible.
53
54
### Production-grade engineering
55
-
- Use `miette` for rich, diagnostic-driven error reporting.
56
- Implement exhaustive integration tests that simulate full backfill cycles.
57
- Adhere to lowercase comments and sentence case in documentation.
58
- Avoid unnecessary comments if the code is self-documenting.
···
60
### Storage and serialization
61
- **State**: Use `rmp-serde` (MessagePack) for all internal state (`RepoState`, `ErrorState`, `StoredEvent`).
62
- **Blocks**: Store IPLD blocks as raw DAG-CBOR bytes in the CAS. This avoids expensive transcoding and allows direct serving of block content.
63
-
- **Cursors**: Store cursors as plain UTF-8 strings for visibility and manual debugging.
64
- **Keyspaces**: Use the `keys.rs` module to maintain consistent composite key formats.
65
66
## Database schema (keyspaces)
67
68
Hydrant uses multiple `fjall` keyspaces:
69
- `repos`: Maps `{DID}` -> `RepoState` (MessagePack).
70
-
- `records`: Maps `{DID}\x00{Collection}\x00{RKey}` -> `{CID}` (String).
71
- `blocks`: Maps `{CID}` -> `Block Data` (Raw CBOR).
72
- `events`: Maps `{ID}` (u64) -> `StoredEvent` (MessagePack). This is the source for the JSON stream API.
73
-
- `cursors`: Maps `firehose_cursor` or `crawler_cursor` -> `Value` (String).
74
- `pending`: Index of DIDs awaiting backfill.
75
-
- `errors`: Maps `{DID}` -> `ErrorState` (MessagePack) for retry logic.
76
-
- `buffer`: Maps `{DID}\x00{SEQ}` -> `Buffered Commit` (MessagePack).
77
78
## Safe commands
79
80
-
### Compilation and linting
81
-
- `cargo check` - fast validation of changes.
82
-
- `cargo clippy` - ensure idiomatic Rust code.
83
-
84
### Testing
85
- `nu tests/repo_sync_integrity.nu` - Runs the full integration test suite using Nushell. This builds the binary, starts a temporary instance, performs a backfill against a real PDS, and verifies record integrity.
86
- `nu tests/stream_test.nu` - Tests WebSocket streaming functionality. Verifies both live event streaming during backfill and historical replay with cursor.
87
- `nu tests/authenticated_stream_test.nu` - Tests authenticated event streaming. Verifies that create, update, and delete actions on a real account are correctly streamed by Hydrant in the correct order. Requires `TEST_REPO` and `TEST_PASSWORD` in `.env`.
88
89
## Rust code style
90
···
22
23
## Project overview
24
25
+
Hydrant is an AT Protocol indexer built on the `fjall` database. It supports both full-network indexing and filtered indexing (eg. by DID).
26
27
Key design goals:
28
- Ingestion via the `fjall` storage engine.
29
- Content-Addressable Storage (CAS) for IPLD blocks.
30
- Reliable backfill mechanism with buffered live-event replay.
31
- Efficient binary storage using MessagePack (`rmp-serde`).
32
+
- Uses `jacquard` suite of ATProto crates.
33
34
## System architecture
35
36
+
Hydrant consists of several components:
37
+
- **[`hydrant::ingest::firehose`]**: Connects to an upstream Firehose (Relay) and filters events. It manages the transition between discovery and synchronization.
38
+
- **[`hydrant::ingest::worker`]**: Processes buffered Firehose messages concurrently. Verifies signatures, updates repository state, detects gaps for backfill, and persists records.
39
+
- **[`hydrant::crawler`]**: Periodically enumerates the network via `com.atproto.sync.listRepos` to discover new repositories when in full-network mode.
40
+
- **[`hydrant::backfill`]**: A dedicated worker that fetches full repository CAR files from PDS instances when a new repo is detected.
41
+
- **[`hydrant::api`]**: An Axum-based XRPC server implementing repository read methods (`getRecord`, `listRecords`) and system stats. It also provides a event stream API via WebSockets.
42
+
- **Persistence worker** (in `src/main.rs`): Manages periodic background flushes of the LSM-tree and cursor state.
43
44
### Lazy event inflation
45
+
46
To minimize latency in `apply_commit` and the backfill worker, events are stored in a compact `StoredEvent` format. The expansion into full TAP-compatible JSON (including fetching record content from the CAS and DAG-CBOR parsing) is performed lazily within the WebSocket stream handler.
47
48
## General conventions
···
54
- Prefer compile-time guarantees over runtime checks where possible.
55
56
### Production-grade engineering
57
+
- Use `miette` for diagnostic-driven error reporting.
58
- Implement exhaustive integration tests that simulate full backfill cycles.
59
- Adhere to lowercase comments and sentence case in documentation.
60
- Avoid unnecessary comments if the code is self-documenting.
···
62
### Storage and serialization
63
- **State**: Use `rmp-serde` (MessagePack) for all internal state (`RepoState`, `ErrorState`, `StoredEvent`).
64
- **Blocks**: Store IPLD blocks as raw DAG-CBOR bytes in the CAS. This avoids expensive transcoding and allows direct serving of block content.
65
+
- **Cursors**: Store cursors as big-endian bytes (`u64`/`i64`).
66
- **Keyspaces**: Use the `keys.rs` module to maintain consistent composite key formats.
67
68
## Database schema (keyspaces)
69
70
Hydrant uses multiple `fjall` keyspaces:
71
- `repos`: Maps `{DID}` -> `RepoState` (MessagePack).
72
+
- `records`: Maps `{DID}|{Collection}|{RKey}` -> `{CID}` (String).
73
- `blocks`: Maps `{CID}` -> `Block Data` (Raw CBOR).
74
- `events`: Maps `{ID}` (u64) -> `StoredEvent` (MessagePack). This is the source for the JSON stream API.
75
+
- `cursors`: Maps `firehose_cursor` or `crawler_cursor` -> `Value` (u64/i64 BE Bytes).
76
- `pending`: Index of DIDs awaiting backfill.
77
+
- `resync`: Maps `{DID}` -> `ResyncState` (MessagePack) for retry logic/tombstones.
78
+
- `counts`: Maps `k|{NAME}` or `r|{DID}|{COL}` -> `Count` (u64 BE Bytes).
79
80
## Safe commands
81
82
### Testing
83
- `nu tests/repo_sync_integrity.nu` - Runs the full integration test suite using Nushell. This builds the binary, starts a temporary instance, performs a backfill against a real PDS, and verifies record integrity.
84
- `nu tests/stream_test.nu` - Tests WebSocket streaming functionality. Verifies both live event streaming during backfill and historical replay with cursor.
85
- `nu tests/authenticated_stream_test.nu` - Tests authenticated event streaming. Verifies that create, update, and delete actions on a real account are correctly streamed by Hydrant in the correct order. Requires `TEST_REPO` and `TEST_PASSWORD` in `.env`.
86
+
- `nu tests/debug_endpoints.nu` - Tests debug/introspection endpoints (`/debug/iter`, `/debug/get`) and verifies DB content and serialization.
87
88
## Rust code style
89