citadeldb

Local-first encrypted memory for AI agents. Raw conversation turns are stored as written, with no summarizer LLM in the ingest path. Citadel runs inside your process; the database uses a companion .citadel-keys file for its encryption keys.

Install

pip install citadeldb

Requires Python 3.10 or later. The only runtime dependency is NumPy; embeddings are bring-your-own. CandleEmbedder and CrossEncoder are not included in the default wheel.

To build from source, activate a virtual environment and run these commands from the repository root with Rust installed:

pip install maturin
maturin develop --release

Semantic embeddings

The default wheel can use a local Sentence Transformers model through Citadel's cancellation-aware embedder protocol:

pip install sentence-transformers

from sentence_transformers import SentenceTransformer


class LocalEmbedder:
    metric = "cosine"

    def __init__(self):
        name = "sentence-transformers/all-MiniLM-L6-v2"
        revision = "1110a243fdf4706b3f48f1d95db1a4f5529b4d41"
        self.model = SentenceTransformer(name, revision=revision, device="cpu")
        self.dim = self.model.get_sentence_embedding_dimension()
        self.model_id = f"{name}@{revision}:sentence-transformers:normalized:v1"

    def embed_with_cancel(self, texts, cancel_token):
        vectors = []
        for start in range(0, len(texts), 16):
            if cancel_token is not None:
                cancel_token.check()
            vectors.extend(self.model.encode(
                texts[start:start + 16], normalize_embeddings=True,
                show_progress_bar=False,
            ).tolist())
            if cancel_token is not None:
                cancel_token.check()
        return vectors


embedder = LocalEmbedder()

The model is downloaded on first construction and runs locally. Keep the model revision and encoding settings unchanged when reopening a region. This wrapper checks cancellation between batches, not during a batch's model inference.

Custom embedders require dim, metric, model_id, and embed_with_cancel(texts, cancel_token). Return one dim-wide vector per input and accept None as the token. Asymmetric models should also implement embed_queries_with_cancel for query-specific encoding. MockEmbedder is only a deterministic lexical test backend, not a semantic model.

Memory

Use the embedder from the example above:

import citadeldb

db = citadeldb.connect("memory.cdl", key="your-passphrase", region_keys=True)
mem = db.memory()
mem.create_encrypted_region("chat", embedder)

mochi = mem.remember("chat", {"kind": "fact", "text": "Alice's cat is named Mochi"})
berlin = mem.remember("chat", {"kind": "fact", "text": "Alice lives in Berlin"})
mem.link("chat", berlin, mochi, "refines")

for hit in mem.recall("chat", text="Where does Alice live?", k=2):
    assert hit.relevance is not None
    print(f"{hit.relevance:.3f}  {hit.text}")

for edge in mem.fetch_edges("chat", src=berlin, limit=100):
    print(edge["kind"], edge["dst"])

Recall fuses vector similarity, keyword match, recency, and importance. Stored text is not summarized or rewritten. AtomHit.relevance is a higher-is-better ranking score; distance is the metric distance when available. Either can be None on hits returned by operations that do not calculate that value.

mem.recall_mmr("chat", text="Alice", k=2, fetch_k=10, lambda_mult=0.5) selects diverse results using the stored candidate vectors, without re-embedding documents. MMR requires a cosine region.

Edges are region-scoped: both endpoints must be live in the named region. Kind summaries are bounded pages; pass next_after_kind back as after_kind until it is None. mem.profile(...) returns recalled atoms together with their induced region-local edges, and mem.unlink(...) removes one exact typed edge.

Local Candle models

To use the Rust-native embedder and cross-encoder in Python, activate a virtual environment and run these commands from the repository root. Building requires Rust:

pip install maturin
maturin develop --release --features candle-embed
pip install citadeldb-mcp
citadeldb-mcp pull e5-large
citadeldb-mcp pull ms-marco-minilm

Use the snapshot directories printed by pull below. For a new region, replace LocalEmbedder() in the memory example with this CandleEmbedder, then set the reranker on mem. An existing region requires re-embedding to change its model.

embedder = citadeldb.CandleEmbedder("/path/to/e5-large", preset="e5-large")
mem.set_reranker(citadeldb.CrossEncoder("/path/to/ms-marco-minilm"))

CandleEmbedder.model_id binds the model, tokenizer, configuration, preset, and embedding pipeline revision. These local directories are user-trusted inputs.

Inspection and model changes

Inspect stored memory without loading an embedder:

maintenance = db.memory_maintenance()
for item in maintenance.inventory():
    print(item.region.name, item.region.model_id, item.live_atoms, item.unavailable)

MemoryMaintenance can inventory, fetch, verify, and forget existing atoms. It cannot remember or recall.

When a model changes, mem.reembed_region("chat", embedder) recomputes vectors from stored text while preserving atom ids and authored edges. Managed similarity edges are rebuilt. It refuses vector-bearing atoms whose original text was not stored.

For a provenance-only repair, use mem.reclassify_region("chat", embedder.model_id) and then mem.attach_existing_region("chat", embedder). Do this only if the model, artifacts, and encoding settings are exactly those that produced the stored vectors: reclassification neither verifies nor recomputes them.

Forgetting

On encrypted regions, forgetting destroys the key that sealed each erased atom. It does not erase plaintext already exported or copies of keys held in pre-erasure backups or snapshots. Plaintext regions do not provide per-atom cryptographic erasure. Targeted forgetting skips immutable atoms unless force=True; their ids appear in receipt.immutable_skipped.

receipt = mem.forget("chat", [berlin])
print(receipt.cryptographic_erasure, receipt.algorithm)
# True AES-256-KW(RFC3394)

Pass cascade_dependents=True to erase the selected atoms together with their transitive derived_from dependents. Without force=True, an immutable atom in that closure rejects the entire cascade. The default remains targeted deletion.

SQL and vector search

The same file is a full SQL database with JSON, full-text search, and filtered vector search.

db.execute("CREATE TABLE notes(id INTEGER PRIMARY KEY, body TEXT)")
db.execute("INSERT INTO notes VALUES (1, $1)", ["hello"])
print(db.query("SELECT body FROM notes").to_dicts())  # [{'body': 'hello'}]

Long SQL, integrity, and memory operations can be stopped cooperatively from another Python thread. Tokens are one-shot; install a fresh one for each unit of work and clear it afterwards.

token = citadeldb.CancelToken()
db.set_cancel(token)
# another thread may call token.cancel()

try:
    report = db.integrity_check(quiet=True)
    for finding in report["errors"]:
        print(finding["kind"], finding["message"], finding["tampered"])
finally:
    db.set_cancel(None)

MCP

pip install citadeldb-mcp installs the server executable for Claude Desktop, Cursor, and other Model Context Protocol clients. The recommended setup is:

citadeldb-mcp pull e5-large
citadeldb-mcp pull ms-marco-minilm
citadeldb-mcp --db memory.cdl --embedder e5-large --reranker ms-marco-minilm

Set CITADEL_KEY to the vault passphrase before serving.

Full documentation is at citadeldb.dev; source and the Rust API are in the main repository.

License

Apache-2.0

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

citadeldb-2.3.0.tar.gz (2.0 MB view details)

Uploaded Sep 7, 2026 Source

Built Distributions

If you're not sure about the file name format, learn more about wheel file names.

The dropdown lists show the available interpreters, ABIs, and platforms. Enable javascript to be able to filter the list of wheel files.

citadeldb-2.3.0-pp311-pypy311_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (8.4 MB view details)

Uploaded Sep 7, 2026 PyPymanylinux: glibc 2.17+ x86-64

citadeldb-2.3.0-pp311-pypy311_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (7.8 MB view details)

Uploaded Sep 7, 2026 PyPymanylinux: glibc 2.17+ ARM64

citadeldb-2.3.0-cp315-cp315t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (8.4 MB view details)

Uploaded Sep 7, 2026 CPython 3.15tmanylinux: glibc 2.17+ x86-64

citadeldb-2.3.0-cp315-cp315t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (7.8 MB view details)

Uploaded Sep 7, 2026 CPython 3.15tmanylinux: glibc 2.17+ ARM64

citadeldb-2.3.0-cp315-cp315-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (8.4 MB view details)

Uploaded Sep 7, 2026 CPython 3.15manylinux: glibc 2.17+ x86-64

citadeldb-2.3.0-cp315-cp315-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (7.8 MB view details)

Uploaded Sep 7, 2026 CPython 3.15manylinux: glibc 2.17+ ARM64

citadeldb-2.3.0-cp314-cp314t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (8.4 MB view details)

Uploaded Sep 7, 2026 CPython 3.14tmanylinux: glibc 2.17+ x86-64

citadeldb-2.3.0-cp314-cp314t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (7.8 MB view details)

Uploaded Sep 7, 2026 CPython 3.14tmanylinux: glibc 2.17+ ARM64

citadeldb-2.3.0-cp314-cp314-win_amd64.whl (8.8 MB view details)

Uploaded Sep 7, 2026 CPython 3.14Windows x86-64

citadeldb-2.3.0-cp314-cp314-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (8.4 MB view details)

Uploaded Sep 7, 2026 CPython 3.14manylinux: glibc 2.17+ x86-64

citadeldb-2.3.0-cp314-cp314-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (7.8 MB view details)

Uploaded Sep 7, 2026 CPython 3.14manylinux: glibc 2.17+ ARM64

citadeldb-2.3.0-cp314-cp314-macosx_11_0_arm64.whl (7.5 MB view details)

Uploaded Sep 7, 2026 CPython 3.14macOS 11.0+ ARM64

citadeldb-2.3.0-cp314-cp314-macosx_10_12_x86_64.whl (8.2 MB view details)

Uploaded Sep 7, 2026 CPython 3.14macOS 10.12+ x86-64

citadeldb-2.3.0-cp313-cp313-win_amd64.whl (8.8 MB view details)

Uploaded Sep 7, 2026 CPython 3.13Windows x86-64

citadeldb-2.3.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (8.4 MB view details)

Uploaded Sep 7, 2026 CPython 3.13manylinux: glibc 2.17+ x86-64

citadeldb-2.3.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (7.8 MB view details)

Uploaded Sep 7, 2026 CPython 3.13manylinux: glibc 2.17+ ARM64

citadeldb-2.3.0-cp313-cp313-macosx_11_0_arm64.whl (7.5 MB view details)

Uploaded Sep 7, 2026 CPython 3.13macOS 11.0+ ARM64

citadeldb-2.3.0-cp313-cp313-macosx_10_12_x86_64.whl (8.2 MB view details)

Uploaded Sep 7, 2026 CPython 3.13macOS 10.12+ x86-64

citadeldb-2.3.0-cp312-cp312-win_amd64.whl (8.8 MB view details)

Uploaded Sep 7, 2026 CPython 3.12Windows x86-64

citadeldb-2.3.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (8.4 MB view details)

Uploaded Sep 7, 2026 CPython 3.12manylinux: glibc 2.17+ x86-64

citadeldb-2.3.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (7.8 MB view details)

Uploaded Sep 7, 2026 CPython 3.12manylinux: glibc 2.17+ ARM64

citadeldb-2.3.0-cp312-cp312-macosx_11_0_arm64.whl (7.5 MB view details)

Uploaded Sep 7, 2026 CPython 3.12macOS 11.0+ ARM64

citadeldb-2.3.0-cp312-cp312-macosx_10_12_x86_64.whl (8.2 MB view details)

Uploaded Sep 7, 2026 CPython 3.12macOS 10.12+ x86-64

citadeldb-2.3.0-cp311-cp311-win_amd64.whl (8.8 MB view details)

Uploaded Sep 7, 2026 CPython 3.11Windows x86-64

citadeldb-2.3.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (8.4 MB view details)

Uploaded Sep 7, 2026 CPython 3.11manylinux: glibc 2.17+ x86-64

citadeldb-2.3.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (7.8 MB view details)

Uploaded Sep 7, 2026 CPython 3.11manylinux: glibc 2.17+ ARM64

citadeldb-2.3.0-cp311-cp311-macosx_11_0_arm64.whl (7.6 MB view details)

Uploaded Sep 7, 2026 CPython 3.11macOS 11.0+ ARM64

citadeldb-2.3.0-cp311-cp311-macosx_10_12_x86_64.whl (8.2 MB view details)

Uploaded Sep 7, 2026 CPython 3.11macOS 10.12+ x86-64

citadeldb-2.3.0-cp310-cp310-win_amd64.whl (8.8 MB view details)

Uploaded Sep 7, 2026 CPython 3.10Windows x86-64

citadeldb-2.3.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (8.4 MB view details)

Uploaded Sep 7, 2026 CPython 3.10manylinux: glibc 2.17+ x86-64

citadeldb-2.3.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (7.8 MB view details)

Uploaded Sep 7, 2026 CPython 3.10manylinux: glibc 2.17+ ARM64

citadeldb-2.3.0-cp310-cp310-macosx_11_0_arm64.whl (7.6 MB view details)

Uploaded Sep 7, 2026 CPython 3.10macOS 11.0+ ARM64

citadeldb-2.3.0-cp310-cp310-macosx_10_12_x86_64.whl (8.2 MB view details)

Uploaded Sep 7, 2026 CPython 3.10macOS 10.12+ x86-64

File details

Details for the file citadeldb-2.3.0.tar.gz.

File metadata

Download URL: citadeldb-2.3.0.tar.gz
Upload date: Sep 7, 2026
Size: 2.0 MB
Tags: Source
Uploaded using Trusted Publishing? Yes
Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for citadeldb-2.3.0.tar.gz
Algorithm	Hash digest
SHA256	`eba1cfb9910d8f56e248f8cae956ccf5d08c3a8708f43e373f1d550310c290c0`
MD5	`3ad9f129291aaa53dca51d489ab6a179`
BLAKE2b-256	`09e982c973dcc722aae0d5e28603f00509c45589e1f685a0900614e299faea7f`

Algorithm	Hash digest
SHA256	`ca0c24e5bfdfb24e7f95519cfa17e97f05bb89c26e2843c2e3e8a5d80140cecc`
MD5	`b2723c251b9e37096c19fcb62c7dd557`
BLAKE2b-256	`8cb65b042e1f10a796e1aa114771941c0391373a5bc63e9ea4c3acfee992628a`

Algorithm	Hash digest
SHA256	`293e83d44fc56393d283604ea986c78c87766d9ce4d6a6df0a4dfdc37186dee3`
MD5	`e2a68c25d3c0e94ed207069f21e11aec`
BLAKE2b-256	`f63d9f9394b4a8720bc8468c94dab7de266a65540f508c855b88aab68cb1792b`

Algorithm	Hash digest
SHA256	`a454aa6b7e76876137c1020c6f0b8420bd6201fa0157e40456af97992f7a78e5`
MD5	`1aaf392ed09247b76a41cde47dfea3d1`
BLAKE2b-256	`023cebee0e06b507126ff9c426092c4d431d2ae0d1c1fd874bd81307fdd05343`

Algorithm	Hash digest
SHA256	`d3e4958381502e503046e275dd17e9706a9741e8b65a3a3685bb0ab7f5dc33dc`
MD5	`fe3fdc2a92a3230e1a7ab3b879932fef`
BLAKE2b-256	`73902d8eb8b0627ef7cf52cf8523d54801defd82923a10ec3cf1b7db3f3fc575`

Algorithm	Hash digest
SHA256	`1a18ccdde7a93dd67aed7e0c2bb5e8342233d8ba6c4a79b96325076ed569e217`
MD5	`fb656b5369608f57ead08b2692c1454b`
BLAKE2b-256	`6665837c0b307b91c5b36507cb2d1c029b4436cf9aeb35eea1bbc47c8a189c61`

Algorithm	Hash digest
SHA256	`9616068387f9f7f39e93978bd493dbd8ce004a77f342d36cbdc29dbe2e4f074f`
MD5	`c8b370364e4976c5438f401be4d1624f`
BLAKE2b-256	`da6e4152b12f9912c67d55029a90bd50377f3d850010f84043f599d5c29fc382`

Algorithm	Hash digest
SHA256	`b730c6909a516edcec6dfe949eb04f97f7c466b8ec82a56bb0584b8cdab4a012`
MD5	`8ea201bb95661dd2a71734f56e9d28ce`
BLAKE2b-256	`eb8bda9ad13a2630128eb920a5cc4dd2c26ef442154257c8396bad5af9325540`

Algorithm	Hash digest
SHA256	`b3c17c83ecccf2557708999848d40aff78b43db42f76f9f1574e0963628ea742`
MD5	`2d0f33047926c33920bb0b1e6006a124`
BLAKE2b-256	`71d33d1d538ef921f3b1db2ee0b998083533661ed3f0d47acce5d3dd14aa0ec2`

Algorithm	Hash digest
SHA256	`18cac6e091ecbb3e9b518847bccb37b1722a96f5199a85d842f5f3ac82ca416d`
MD5	`d013336769ac5348874a67b85af98349`
BLAKE2b-256	`b608bd32ebb8b39eb8784c69410c4b47d05233d5e950021fb492e70664bad509`

Algorithm	Hash digest
SHA256	`950e9b4e06bb0611ca0c3b1151389704ab396cb4107a50a3a74ad4cf1624948c`
MD5	`3bad2cf7378a8136317def65325cfe28`
BLAKE2b-256	`20ab9e7e0b26c526cf9dc982aec66a417cb0b98562e38b8478f2d7a876eb4cd1`

Algorithm	Hash digest
SHA256	`c235297f80f010660dc1771c69954844dd720c44c4aaf942b9b3895772b8171f`
MD5	`e783bcf2cf36e59267b4eaffc51ef56d`
BLAKE2b-256	`229b02344ea858eccb420fa1d8d2162051c0e626e88ff6b14a8f2c23ace3083b`

Algorithm	Hash digest
SHA256	`40e84e023d105031d0729c5284bf5bb02d2da51e3dd5c55f2925dded2548bef4`
MD5	`04fcaf282eece77e4f61eb3c0b2bde2b`
BLAKE2b-256	`208a2102480152809aa289a3e06f98ecb4c5d8e084e8bbef0ef19f501c9c379c`

Algorithm	Hash digest
SHA256	`322070a9e1fc436dda421192b5ace54438d08fdea4279363c16e62ae248607a7`
MD5	`ec9e9c452ce256a1ecdc081109086d5e`
BLAKE2b-256	`78700d731280bb250aed0f242a6faa5d3d6ca6c7aa600a7b5059525168d5fced`

Algorithm	Hash digest
SHA256	`ab22c81ee4521d75d09856b72e89927b28dce39b466c6246ab98313fda36ff39`
MD5	`b7680392ca367f13938d98e0f993b0aa`
BLAKE2b-256	`2974c469649962a5ca3e2444d8586dc0c5716dcdc7e006af066ec7efc61728b2`

Algorithm	Hash digest
SHA256	`e6cd9f90e9495c21800b8e6d651bbf2fe7ca8f7b9ae5c23ee2f35d9ce1a8774a`
MD5	`461f5025630a9ee3ee8d8d9fd715d1b6`
BLAKE2b-256	`9c09bfd237e63b1369792382cedcebb44c11dded7511b12d479be88925992e8b`

Algorithm	Hash digest
SHA256	`83b91efa14a7c87df9c1eed77de781f6458543341a0cbe935e9026c81040ad72`
MD5	`799703844c4da46ad679f4f528629b84`
BLAKE2b-256	`7904dc4d1a041e8f5032afbc60409bc66da8c5f86cfedbfaff35bc5f63a2fab7`

Algorithm	Hash digest
SHA256	`7df9c6c1f921d400f1c639efe4911f614be14b4c918ed40da9a909f54aedbe28`
MD5	`fc9800d5010c955489fb6769dbcdddcf`
BLAKE2b-256	`5c4b4a5721d5686913e17b36822649fed400e0baa63a4d4bc12af902a0f47d9a`

Algorithm	Hash digest
SHA256	`d6fe1768eaeab6289fe1e24286f9f8c9867bc2977bfc0d46e0f46d77763a9e8b`
MD5	`0c95582e0e3dd6c66a893744860e089a`
BLAKE2b-256	`4600c36078a101823c2f02662e8b24035a294d4b8fd48c4f80d20c5b7933de8a`

Algorithm	Hash digest
SHA256	`8e061b6f85d042e76a3bc948e3006ff58032a59e2889bdd17cd2908f0c4a7df0`
MD5	`23e5501764cce4790b3cf493a18ae8e7`
BLAKE2b-256	`c694bdca98a6bf872b7672093db1abf6b21797cb2606711c1f09c655b74e2890`

Algorithm	Hash digest
SHA256	`43d30c4b2731beb315d8d609065ec6108aeb7644f8cb4a556d7d70eecb304994`
MD5	`02301e6acdb6f44e36abdea0a8bdcc8d`
BLAKE2b-256	`2e1d9bf5790396c3fe0dc4d4aebc79ec051e7c5f7c0eebb8b0afa5cdea8e6ab4`

citadeldb 2.3.0

citadeldb

Install

Semantic embeddings

Memory

Local Candle models

Inspection and model changes

Forgetting

SQL and vector search

MCP

License

Project links

Key dates

1 maintainer

Credits

GitHub Statistics

License expression

Requires

Tags

Classifiers

Download files

Source Distribution

Built Distributions

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

Algorithm	Hash digest
SHA256	`7e980a94039bc0b392b3365730b04710db695b8574081924e5fb4c4dc866e997`
MD5	`06ed5e0ba24c98612294d1f581c65d95`
BLAKE2b-256	`58612b028d6ec1778037868f8b5e55c43ca020788b53138e2d7fad7f04304537`

Algorithm	Hash digest
SHA256	`1cb379937d0790328cddf45729ca447c5721b9e3bd66a4880e5e5796d84458f0`
MD5	`f3c723742a7fd1e7f83569867865d502`
BLAKE2b-256	`818bfec6f56f3d00f775ff2f0e66ea4ec37cabc11ca94db36ba29a406fd29cf8`

Algorithm	Hash digest
SHA256	`bf9c33d50b6f6e7c3364345f09d8cff2750f99a454a0fb5924271be5b2c29f31`
MD5	`b27f9ab239cae667881c2e0836871b58`
BLAKE2b-256	`7812c57e7b65688354cfe31561e1f4e66ede95b8e5ace10392224674eceb7a8c`

Algorithm	Hash digest
SHA256	`0fd81fcf35a8e15ec99422c62ca8697e9e630c628c5aa6eb980e7b6f70f38b06`
MD5	`23159effbd33b816f3e8f79cae1a9e26`
BLAKE2b-256	`8c57027a7c740a6be449d1d0a628ed2efb195812a91e207f6274e2c994e1cd90`

Algorithm	Hash digest
SHA256	`cf1372ceb9de7415587dc0716319f4740ae6391ce27af67b4cf5e84fb3b0a659`
MD5	`a4ece54a29e5e1c2af24717eb95ae262`
BLAKE2b-256	`021d452238ea7695c4760b4ff1856f91bed0fd8f49a9368a2c75629892e9c927`

Algorithm	Hash digest
SHA256	`fa777beabbe1e49d2578c01041e897dff12cb354db68326add71fc6ad24c11c6`
MD5	`09dae45aed2751017db2cd085ec313a5`
BLAKE2b-256	`ad716afef3e83fe405db0e7e69a5c0b57ca76c2a2888efd4dce7c062e59ff106`

Algorithm	Hash digest
SHA256	`1bf4198a129a9977fe5805ee9ce9cc19fa0ab8c81145b725ad075c1ac0f3fbfd`
MD5	`f3c4d7eb8f8eded865d5e73f82323ed4`
BLAKE2b-256	`01d84bde8b9bac56bb425994fdd93bc5c53f9baa97911f9f2ea772045974ad06`

Algorithm	Hash digest
SHA256	`dcb09d43b2cd92c0110e8320330551f7a1ce45c357f04dedb121f2d6190f6561`
MD5	`9e5a2bf20f506a0a7e46ab07fef450cb`
BLAKE2b-256	`8749ce675b4b0cd86260710c1b7c9ad0c17bbc80a2f198d2980ee1797595efaa`

Algorithm	Hash digest
SHA256	`aa6b94c746b96f2d99a30d4316b7349311c84aa7bcbb66c692ff6fb24fffa275`
MD5	`76d536e3e052796b7ae58a403605ad0e`
BLAKE2b-256	`4e25efe8cc7767b6e2cee864e85eced3da61fdddecbc76f91f89a61e958caade`

Algorithm	Hash digest
SHA256	`4f1d2693b508b4291317327076b4710e25353b6e3864d2b13ce18d1705dc3d46`
MD5	`466af3a4db0a14eb756c5cc2e7941826`
BLAKE2b-256	`071103db72b27d563f0d2d8bccd243f8c2f4b0f1149139eb40da6fc102f464b6`