Kingfisher Secret Scanner for Python
Fast secret detection, credential validation and revocation for Python, powered by Kingfisher's Rust libraries. Native, in-process execution; no CLI subprocess required. Requires CPython 3.10+.
Install
uv add kingfisher-secret-scanner
The PyPI distribution is kingfisher-secret-scanner; the Python import is
kingfisher_sdk. It coexists with the separately distributed kingfisher-bin CLI.
The prepared SDK version is 1.6.3, a backward-compatible patch release.
The SDK version is independent of the CLI and Rust library versions.
SDK/runtime changes need a new PyPI version. The release workflow checks this before native builds, while default manual runs remain test-only. Maintainers can check the prepared version without building or publishing:
python3 scripts/check-sdk-release.py --require-new --expected-version 1.6.3
Explicit publication requires upstream main, publish=true, and matching
expected_sha / expected_version safety inputs on the existing workflow. See
the publishing guide for the
exact dispatch command, six-wheel/sdist test gates and failed-upload recovery.
CLI and low-level Rust validation responses can include
allowlisted identity metadata
without extra provider requests. Python ValidationResult representations and
serialization still omit provider responses and private validation evidence;
use the explicit analysis API when you need identity or permission details.
Scan offline
from kingfisher_sdk import Scanner
scanner = Scanner() # Compile the bundled rules once and reuse the scanner.
for finding in scanner.scan_file("application.conf"):
if finding.visible:
print(finding.to_dict()) # Secrets and captures are redacted by default.
Inspect rules, regexes and configured actions
Custom YAML HTTP verification services can use
OAuth2 authentication.
Reuse a Validator for token reuse, expiry refresh and failure cooldowns; set
credential environment variables through the rule's *_env fields.
HTTP validation uses only the primary endpoint by default. To explicitly permit up
to two rule-defined fallback environments, use Validator(http_fallbacks=True).
reuse_primary_validation_logic defaults to false; true reuses primary HTTP validation
logic with its URL explicitly replaced. reuse_primary_revocation_logic separately opts in to primary
revocation reuse with replacement URL(s); false/omitted requires an independent
action or leaves revocation unavailable. The CLI flag is --fallbacks for both
capabilities; revocation never fails over mutation targets. Existing rules
need no edits. See the rule guide
and runnable example.
Neither switch inherits primary URL(s).
CLI revoke --fallbacks validates to select a target automatically. Python
result-based revocation keeps its existing contract: reuse the retained snapshot
with Revoker.revoke(result, confirm=True), without another validation request.
Exact-ID revocation uses only the primary action and rejects reserved VALIDATION_*
endpoint-selection variables; pass a validated result to revoke a fallback credential.
import json
from kingfisher_sdk import Rules
rules = Rules(confidence="low") # Include the complete bundled catalog.
print([r["id"] for r in rules.metadata() if r["revocation"]])
detail = rules.detail("betterleaks.aws-access-token") # Exact loaded rule ID.
print(detail["pattern"])
print(detail["detection_regex"]) # Compiled confirmation regex, comments removed.
print(json.dumps(detail["validation"], indent=2))
print(json.dumps(detail["revocation"], indent=2))
detail() also includes entropy, path/candidate filters, capture selection,
dependencies, examples and references. It returns a copy; edits do not change
scanning. Absent actions are None. HTTP/gRPC configurations show requests and
matchers; Betterleaks logic is a serialized expression tree. Typed/raw handlers
identify Rust implementations rather than exposing their source code. Inspection
makes no provider requests and does not execute revocation. Literal OAuth2 usernames,
passwords and client secrets are redacted; their *_env names remain visible without
resolving environment variables. Other custom rule literals are returned as configured.
Inspection never changes the credentials used by later explicit validation.
The rule example supports catalog capability filters and detailed field selection:
uv run --no-sync python python/examples/rules.py --with-revocation
uv run --no-sync python python/examples/rules.py betterleaks.aws-access-token --field pattern --field validation --field revocation
Filter findings and choose inputs
The shared built-in catalog includes betterleaks.generic-api-key at static low confidence.
Default Rules() and Scanner() use minimum medium confidence, excluding it before native
compilation, as does the default CLI.
Opt in with Rules(confidence="low") and pass those rules to Scanner(rules).
betterleaks.generic-credential-uri stays medium; generic-password and generic-username
remain excluded from the catalog. Generic rules are not all low confidence.
All rules use configured confidence only; filters cannot upgrade or downgrade findings,
and there are no runtime confidence variants. Custom YAML/TOML files still load without
edits: legacy one-string setConfidence(...) and filter.setConfidence(...) calls are
inert value passthroughs. Rejection filters and credential-association helpers remain active.
Rules() and the default Scanner() reuse the CLI's compiled Vectorscan rule
cache. Use Rules(cache_dir="rule-cache") to choose a directory, or set
KF_RULE_CACHE_DIR; an explicit directory wins. Otherwise the OS cache directory
is used. Rules(cache=False) disables all cache reads and writes.
Prewarm at image build time with the same rules and confidence used at startup.
Build as the runtime account, or transfer both the cache directory and its files
to that account with COPY --chown; ownership is checked on the receiving files,
not tied to the builder's identity. The commented
container example demonstrates root image-build
prewarming followed by a non-root runtime, including read-only cache hits.
Reuse entries with the same wheel/engine build and compatible CPU architecture; native CPU/version rejection or corrupt entries trigger automatic
recompilation. Cache writes are best effort. Older cache formats recompile once.
Rule loading and confirmation/path/filter initialization still run per Rules();
reuse a scanner within each process. The cache stores rules, not scan results. Cache directories must be trusted.
Unix directories are private; unsafe ownership/permissions disable caching, and
no default falls back to shared temporary storage. Windows cache directories/files
must belong to the caller; DACLs also permit Administrators/SYSTEM, with
TrustedInstaller permitted for protected ancestors only. Payload SHA-256 detects
corruption before native deserialization, rather than authenticating cache writers.
See the commented cache example and
cache guide.
Use Rules(confidence="high") to load only high-confidence rules, or
Rules(["company.yml"], builtins=False) for a custom-only YAML/TOML catalog.
Scanner(base64=False) disables Base64 detection. Leave native entropy thresholds
at their defaults unless you intend to override every rule's threshold.
Scanner and DetectionScanner enforce custom YAML pattern_requirements
(character minimums, ignored substrings and Liquid checksums) for raw and Base64
matches. The luhn_valid and iban_valid boolean filters simplify offline
structured-data checksums. Run the commented example
with uv run --no-sync python python/examples/pattern_requirements.py; output stays redacted.
See the rule guide for capture semantics and filter limits.
Findings retain compact full-match component context automatically. Validation
honors within and explicitly supported verify_candidates opt-ins with the
CLI's ranking and bounded search. Identical rule/resolved inputs share one request
per validation batch; different rules and components remain independent.
Scanning stays offline and revocation requires explicit confirmation.
Built-in PEM detectors share bounded offline structure checks with the CLI. BER key encodings, incomplete outer markers and partial diff prefixes are kept when they may contain real key material; these checks do not prove live validity.
For report filtering, serialize once and inspect the redacted fields:
from fnmatch import fnmatchcase
from kingfisher_sdk import Scanner
findings = Scanner().scan_file("application.conf")
for finding in findings:
data = finding.to_dict()
if finding.visible and fnmatchcase(data["rule_id"], "betterleaks.*"):
if data["entropy"] >= 3.5:
print(data)
Native finding properties expose these fields without serializing the whole result. Redaction retains fingerprints, entropy and content identity, which can disclose information about guessable secrets. Keep reports access-controlled.
You can also filter confidence (low/medium/high), Base64 status, line/byte
location, capture-name presence, fingerprint or blob ID. Keep paths alongside
findings for path filters; Finding does not contain a filename. After validation,
filter result.outcome, result.http_status or result.reason. Retain invisible
helper findings until validation finishes. Report filters do not reduce detection
work or validation requests.
The scan example
supports directories, repeated --rules-path, --confidence, --exclude, --rule,
--min-entropy, --base64-only, --no-base64, --unique, per-file --timeout,
and Ctrl-C cancellation. Its entropy option filters output, while
Scanner(min_entropy=...) overrides detector thresholds. --unique deduplicates
reports by rule and credential value using per-run keyed digests internally; SDK dedup=True instead skips repeat successful
scans of identical content at the same path.
uv run --no-sync python python/examples/scan.py ~/example-repo --timeout 5 --unique --exclude '*.min.js'
uv run --no-sync python python/examples/scan.py application.conf --rule 'betterleaks.*' --min-entropy 3.5
uv run --no-sync python python/examples/validate.py application.conf --outcome verified_active --concurrency 4
The directory example skips .git, .venv, node_modules and symlink entries;
its path globs are not .gitignore syntax. The composable input APIs below support native filesystem/Git history enumeration
and archive/content expansion. DetectionScanner opts into shared CLI matching,
bounded nested Base64, inline-ignore and HTML/CSS checks; validation remains explicit. Enumeration and CLI deduplication can change results
compared with kingfisher scan. See the
filtering and CLI comparison guide.
Detection policies, SQLite/bytecode/PDF and Git scopes
from kingfisher_sdk import DetectionPolicy, Scanner, GitScope, git_inputs, expand_archives, expand_content
sources = git_inputs("./repo", scope=GitScope(since_commit="main"))
sources = expand_content(expand_archives(sources, depth=2))
for result in Scanner(policy=DetectionPolicy()).scan_inputs(sources):
for finding in result.findings:
if finding.visible:
print(result.input.path, [origin.id for origin in result.input.origins], finding.to_dict())
Scanner(policy=DetectionPolicy(...)) adds CLI matching, bounded nested Base64,
inline-ignore and HTML/CSS checks. The frozen policy is reusable; the existing
DetectionScanner convenience API constructs that same policy. Default Scanner
behavior is unchanged. It defaults to full-match
component windows, URI fallback/secret containment suppression, two Base64 layers,
and a 64 MiB original-input cap for the Base64 pass. Raw scanning still runs above
the cap; nested findings keep the outer encoded region's offsets.
Containment checks are separate for raw input and each decoded buffer, preserving
separately encoded sibling secrets even when they share those outer offsets.
Set cli_match_semantics=False, base64_max_depth=1, and
base64_max_input_bytes=None to retain legacy SDK matching/decoding while choosing
context filters. Each policy can be configured independently. expand_content extracts SQLite SQL and .pyc strings without
executing bytecode. SQLite lock waits remain bounded independently of content limits.
Use expand_content(sources, pdf=True) to explicitly extract PDF page text as
document.pdf!page-N.txt, preserving provenance and assigning offsets within each
page's UTF-8 text. Input, output, decompression and page-count limits apply; no
OCR, encrypted PDFs, embedded files, annotations, or Form XObject text extraction.
Reported page/font stream decode errors are skipped by default without losing
readable pages; strict=True raises on those errors instead of yielding partial pages.
Neither mode guarantees complete coverage when a codec silently recovers corruption.
See the PDF example
and limits/fallback guide.
GitScope supports history ranges, inclusive branch roots,
time bounds, snapshots, net diffs, staged index content and unreachable objects.
since_hours must be positive and fit in signed 64-bit seconds; scope bounds are
validated when constructing GitScope.
GitInput.origins retains author/committer identities, times, parents and messages
through extraction. Commit metadata is shared across origins, tree comparisons
skip unchanged subtrees, and payloads load lazily. max_blob_size, max_commits
and max_inputs bound Git work; missing objects raise errors unless explicit
skip_missing_blobs=True requests warning-and-skip behavior. Use discover=False
to require a repository root. Non-UTF-8 paths retain their bytes as raw_path.
Keep messages out of logs; they may contain credentials.
Commented examples: detection, SQLite/bytecode, and Git scopes. See the SDK guide for scope semantics, extraction limits, WAL handling and controls.
Compose enumeration and archive scanning
from itertools import chain
from kingfisher_sdk import Scanner, filesystem, git_history, expand_archives
inputs = chain(filesystem(["./repo"]), git_history("./repo", refs=["HEAD"]))
for result in Scanner().scan_inputs(expand_archives(inputs, depth=2)):
for finding in result.findings:
if finding.visible:
print(result.input.path, result.input.commit, finding.to_dict())
Supply your own Python enumerator with ScanInput.from_file(path) or
ScanInput(path="config.txt", data=blob_bytes, commit=commit_hex) instead of either
native iterator. Both use the same scanner and archive transform. Native Git runs
without a Git executable and enumerates reachable file versions, including merge
parents; filesystem enumeration honors local ignore files and skips symlinks.
Archive expansion shares CLI extractors, with explicit depth and per-root budgets.
Repeated TAR member names retain each occurrence's bytes. The byte budget also caps
intermediate decompressed streams (including TAR headers/padding); exhaustion
raises before any members from that root are yielded. Budgeted TAR/ZIP/ASAR
expansion also rejects malformed or unsafe members and staging I/O errors for the
whole root, without yielding partial members. Other shared format extractors retain
their per-format limits and may skip unreadable entries, so success does not
guarantee complete coverage. scan() and scan_file() keep their existing behavior.
Each ScanResult retains its source and complete findings for separate validation.
See the input API guide
for formats, metadata, limits and cooperative controls, and the
composable example
for --filesystem-enumerator python or engine workflows. The commented
Git-history example
lets callers choose native traversal or a Python adapter using Git subprocesses;
only the latter requires Git on PATH. The
archive example
creates synthetic nested ZIPs and compares file/byte inputs, depth and budgets:
uv run --no-sync python python/examples/git_history.py ./repo --enumerator engine --all-refs
uv run --no-sync python python/examples/git_history.py ./repo --enumerator python --ref HEAD
uv run --no-sync python python/examples/archives.py --source bytes --depth 2
uv run --no-sync python python/examples/archives.py --source file --depth 1
Download demo.yml beside archives.py when running outside a checkout. These
examples report redacted results; validation remains a separate explicit stage.
Scan timeouts and cancellation
from kingfisher_sdk import CancellationToken, Scanner
scanner = Scanner()
cancellation = CancellationToken()
findings = scanner.scan("text to inspect", timeout=2.0, cancellation=cancellation)
# Another thread can call cancellation.cancel() while scanning runs.
scan_file accepts the same keyword arguments. Timeout values must be finite,
positive seconds; omit timeout for unlimited execution. An interrupted scan
raises TimeoutError for an expired deadline or RuntimeError for cancellation,
without partial findings or a dedup-cache entry. Tokens remain cancelled; create
a new token for new work.
These controls are cooperative. Checks run between matching/filtering operations, in native match callbacks and during Base64 enumeration. File reads, decoding, rule compilation and individual native operations cannot be preempted. Python releases the GIL during scanning, allowing another thread to signal cancellation. Use a separate process when a hard wall-clock execution limit is required.
Betterleaks source filters can inspect explicit metadata with
scanner.scan_with_attributes(content, {"path": "config.env", "git.author_name": "alice"}).
This additive method preserves the scanner's detection policy, component context,
timeout/cancellation controls and source-aware deduplication. It performs no source
acquisition. See the detection example.
Validate explicitly
SMTP URIs use the shared validator. For separate SMTP fields, the SMTP example loads custom rules and authenticates over TLS without sending email. See the SMTP input contract for port fallback and outcome classification.
SMTP session cleanup has a short independent timeout so stalled cleanup writes finish promptly after an authentication result is established.
from kingfisher_sdk import Scanner, Validator
findings = Scanner().scan_file("application.conf")
for result in Validator(timeout=10, concurrency=4).validate(findings, timeout=30):
if result.finding.visible:
print(result.to_dict())
Scanning is offline. Validation may contact providers using detected credentials.
Keep the complete scan result, including invisible supporting findings, until
validation finishes. Report redacted results; do not log finding.secret.
For Betterleaks rules with an analyze program, explicitly call
validator.analyze(result, timeout=30) after a verified_active result. Analysis
reuses that result's original captures, components and private validation evidence;
it may make additional provider requests. Ordinary validation never runs analysis.
The returned identity, capabilities and metadata can contain sensitive information.
Private handoff evidence is omitted from ValidationResult.to_dict() and repr.
See the runnable analysis example.
Pass a validated result directly to revoke using its original rule and resolved credential pair. No credential environment variables or manual AWS association are needed:
from kingfisher_sdk import Rules, Scanner, Validator, Revoker
rules = Rules()
findings = Scanner(rules).scan_file("application.conf")
revoker = Revoker(rules)
validated = [result for result in Validator().validate(findings) if result.finding.visible]
for status in revoker.revoke_many(validated, confirm=True):
print(status.rule_id, status.outcome)
This can disable or delete real credentials. confirm=True remains mandatory;
validation alone never revokes. Each validation result's credential exposes the selected supporting
findings, validation variables and resolved revocation inputs. Secrets are hidden
in repr and redacted by default in credential.to_dict(). Missing or ambiguous
components, inactive credentials and unsupported rules cannot be revoked through
this result-based API. Keep each input's complete findings until validation finishes.
The existing revoke(rule_id, secret, variables=..., confirm=True) API still works.
Custom YAML revocation: actions work with the same APIs. Supply files or
directories through Rules; omit builtins=False to also load the bundled catalog:
rules = Rules(["company-rules.yml"], builtins=False)
revoker = Revoker(rules)
plan = revoker.prepare_from_file("application.conf", provider="acme")
# Optional live steps after inspecting the plan and selecting intended targets:
# results = plan.validate(Validator())
# statuses = revoker.revoke_many(results, confirm=True)
Single and batch result-based revocation retain the original custom action and
resolved inputs. A rule needs a revocation definition; custom Betterleaks TOML
alone does not supply one. The custom-rule guide
shows endpoint variables and YAML, and the local lifecycle example
runs demo.yml's action against a mock provider. The preparation
and batch examples accept repeated --rules-path and --no-builtins.
For a token obtained elsewhere, prepare candidates without knowing its rule ID:
import os
from kingfisher_sdk import Revoker
revoker = Revoker()
plan = revoker.prepare(os.environ["KINGFISHER_SECRET"], provider="github")
print(plan.candidate_rule_ids) # Offline identification; no provider requests.
results = plan.validate() # Explicit live checks; never revokes.
# Inspect/disambiguate results before revoker.revoke(chosen_result, confirm=True).
provider accepts an exact loaded rule ID, an exact display name, or an ID family.
Exact IDs take precedence; exact names and families are case-insensitive, and
ambiguous display names require an ID. github selects every loaded ID in that
family (github-..., github.... or github/...), while
betterleaks.github-pat selects just that exact rule. Family matching checks full
IDs and IDs without the supported catalog namespace; it excludes substrings like
githubish and does not classify compact IDs such as Veles's
slackappconfigaccesstoken under slack. Rules.resolve() exposes this offline
selection. Typos may get suggestions, but never select fuzzy matches; globs and
CLI legacy aliases are not supported.
Preparation scans the shared catalog and filters findings. Validation checks
only detected revocation-capable candidates using supporting context, retaining
compound credential associations; it does not try every selected provider API.
Helpers aren't separately checked against providers. Preparation retains primary
matches missing required companions for actionable diagnostics; normal scanner
behavior stays unchanged.
All overlapping candidates are returned without choosing a winner. Supply source
text when a bare token lacks context/companions, or use file preparation for path-aware
rules. Omit provider to consider the loaded catalog. Empty candidates do not prove
inactivity. See the selector reference
and commented preparation example.
Source preparation and batch revocation compose with the same validated snapshots:
from kingfisher_sdk import Revoker, Validator
revoker = Revoker()
plan = revoker.prepare_from_file("application.conf", provider="github")
# For in-memory configuration: prepare_from_text(text, path="application.conf").
for candidate in plan.candidates:
for issue in candidate.input_diagnostics:
print(candidate.rule_id, issue.variable, issue.rule_id, issue.message)
results = plan.validate(Validator(timeout=10, concurrency=4))
# After inspecting/disambiguating all targets; this performs real revocations:
statuses = revoker.revoke_many(results, confirm=True, timeout=60)
for status in statuses:
print(status.index, status.rule_id, status.outcome, status.duplicate_of)
Preparation stays offline and exposes capability flags and named missing,
ambiguous, redacted or invalid inputs without values/endpoints/error bodies.
prepare(..., variables=...) retains explicit supporting/endpoint overrides for
inspection and validation; detected captures/components retain precedence.
Required detected companions still need source context. Validator.inspect()
inspects existing complete finding groups offline; with_variables() clones its
execution policy and shared client with explicit overrides. Prepared validation
skips known missing validation inputs before network work. Revocation-only missing
inputs allow validation but block revocation readiness. These diagnostics do not
prove activity or every provider/expression constraint.
revoke_many() requires confirmation, checks the entire iterable's types before
mutation, skips non-ready results and returns one safe status per original input.
It deduplicates exact original action definitions and complete resolved inputs
within the call, preserving different AWS pairs and endpoints. Different rule IDs
remain distinct, so callers must resolve overlapping detector ambiguity first.
Failures before mutation return failed; ambiguous submitted actions return
unknown, including a mutating HTTP 429 response. Rate limiting does not prove the
action was rejected before it was applied. Processing continues without retries; response matcher
failures return not_confirmed, without proving the mutation did not occur.
Total deadline/cancellation/Ctrl-C stops pending actions and propagates an exception
carrying .completed statuses and .pending_indices. Interrupted in-flight actions
are unknown; inspect provider state before retrying. See the
batch reference and the
commented batch example, which stays offline by default.
The six-provider example
scans, validates and optionally revokes AWS, GCP, Hugging Face, GitLab, Slack and
GitHub credentials. The
local lifecycle example
uses a loopback mock and synthetic credentials. Blocking SDK operations release the
GIL; use asyncio.to_thread in async applications.
Documentation and examples
- Python guide on GitHub:
API reference, custom YAML/TOML rules, validation outcomes, revocation, local
uvbuilds and tests, supported platforms, and publishing. - Runnable Python examples: offline scans, filesystem/Git enumeration, nested archives, live validation, explicit revocation and a safe local lifecycle.
- Kingfisher repository: source, project documentation, CLI features and contributions.
- Report an issue.
For a source checkout, build and run the local example from the repository root:
uv sync --locked --no-install-project
uv run --no-sync maturin develop --locked --profile dev
uv run --no-sync python python/examples/local_workflow.py
For a downloaded example and a published package, use:
uv run --no-project --with kingfisher-secret-scanner python local_workflow.py
Download demo.yml beside local_workflow.py; the example loads that synthetic
rule by its location. Source archives include the examples and fixture together.
Provider calls also accept an optional total deadline and CancellationToken:
validator.validate(findings, timeout=30, cancellation=token) and
revoker.revoke(..., confirm=True, timeout=10, cancellation=token). They check
Ctrl-C while waiting, propagate validation errors without partial results, and add no
retries. Interruption cannot undo a revocation already applied by the provider.
Batch revocation preserves completed statuses on interruption exceptions.
Archive/content transforms validate depth and byte/entry options on call while
consuming inputs lazily. Archive max_bytes caps cumulative extracted output;
the root input has an independent cap of the same size. Buffers and Python-byte
copies can coexist, so this is not a total memory limit. Entry accounting follows
the format: TAR/ZIP include directories, ASAR counts indexed files, and HWP counts
streams. filesystem(max_file_size=...) silently skips oversized files.
Metadata
Release files for kingfisher-secret-scanner 1.6.3
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| kingfisher_secret_scanner-1.6.3.tar.gz | 2.0 MB | With provenance | Details for kingfisher_secret_scanner-1.6.3.tar.gz |
Built distributions (wheels)
Total release size: 113.6 MB
Release files / kingfisher_secret_scanner-1.6.3.tar.gz
| Download URL | kingfisher_secret_scanner-1.6.3.tar.gz |
|---|---|
| Size | 2.0 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
c9d1216af1f4db4e3dc1ab742dcdb332f1527fa87ccfc138bfdc5d6fe9d4504a
|
|
BLAKE2b-256 checksum How to use checksums |
0a6b1637a13f095d4a6a414014be898d5dd1c9d1197e347da27c482827d6bffa
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 9, 2026.
Transparency logRelease files / kingfisher_secret_scanner-1.6.3-cp310-abi3-win_arm64.whl
| Download URL | kingfisher_secret_scanner-1.6.3-cp310-abi3-win_arm64.whl |
|---|---|
| Size | 15.7 MB |
| Tags | CPython 3.10 Windows ARM64 abi3 |
|
SHA-256 checksum How to use checksums |
2217ac92322da872f544097ee92c190b033984e193e6d7abfd2b21d76e0f3298
|
|
BLAKE2b-256 checksum How to use checksums |
26617eecc9c80c79f21f6bc44ad09699c052435556e17538a936b21c2341589a
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 9, 2026.
Transparency logRelease files / kingfisher_secret_scanner-1.6.3-cp310-abi3-win_amd64.whl
| Download URL | kingfisher_secret_scanner-1.6.3-cp310-abi3-win_amd64.whl |
|---|---|
| Size | 28.1 MB |
| Tags | CPython 3.10 Windows x86-64 abi3 |
|
SHA-256 checksum How to use checksums |
4657e3644f553c8653535af7aba0fce799b5a7e096cfc9e10fd1d9efcc29151c
|
|
BLAKE2b-256 checksum How to use checksums |
bc40bf9fbf20c6507a6e4c24c03c1224640c138439bc41b15f358cef2072ed59
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 9, 2026.
Transparency logRelease files / kingfisher_secret_scanner-1.6.3-cp310-abi3-manylinux_2_28_x86_64.whl
| Download URL | kingfisher_secret_scanner-1.6.3-cp310-abi3-manylinux_2_28_x86_64.whl |
|---|---|
| Size | 18.1 MB |
| Tags | CPython 3.10 Linux glibc 2.28+ x86-64 abi3 |
|
SHA-256 checksum How to use checksums |
e4606919f8e3964b99a46b94cc1f241750b3688c80f8273a2ab5c3f6e32ff54e
|
|
BLAKE2b-256 checksum How to use checksums |
c5e10456549b4f2468645aa95759045cfcb305607701bc3c13526230ab202061
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 9, 2026.
Transparency logRelease files / kingfisher_secret_scanner-1.6.3-cp310-abi3-manylinux_2_28_aarch64.whl
| Download URL | kingfisher_secret_scanner-1.6.3-cp310-abi3-manylinux_2_28_aarch64.whl |
|---|---|
| Size | 17.4 MB |
| Tags | CPython 3.10 Linux glibc 2.28+ ARM64 abi3 |
|
SHA-256 checksum How to use checksums |
1d7433d9edeb741e7d1b9089e6f271ecf948f558ee9a8a843b18dbdac0278efa
|
|
BLAKE2b-256 checksum How to use checksums |
2f2177500af764e01ab4e63100b401a1bb85c3f1c4b2a415d526265a002dd030
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 9, 2026.
Transparency logRelease files / kingfisher_secret_scanner-1.6.3-cp310-abi3-macosx_11_0_x86_64.whl
| Download URL | kingfisher_secret_scanner-1.6.3-cp310-abi3-macosx_11_0_x86_64.whl |
|---|---|
| Size | 16.4 MB |
| Tags | CPython 3.10 abi3 macOS 11.0+ x86-64 |
|
SHA-256 checksum How to use checksums |
a5182c6916e7e0831c9b24c82816b7271f99accecbc9b5a9c34bb72010a62d3d
|
|
BLAKE2b-256 checksum How to use checksums |
a818383bc1b68ec976d727b58b7f3cca090b52f4b6de4874cbb1d6e8cd24c307
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 9, 2026.
Transparency logRelease files / kingfisher_secret_scanner-1.6.3-cp310-abi3-macosx_11_0_arm64.whl
| Download URL | kingfisher_secret_scanner-1.6.3-cp310-abi3-macosx_11_0_arm64.whl |
|---|---|
| Size | 15.8 MB |
| Tags | CPython 3.10 abi3 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
e60d7f097875906d4c2c54dd768d4a60a8a5bba0475fbe34fa3fb0f8b1ba47b4
|
|
BLAKE2b-256 checksum How to use checksums |
d1ddb63bfe580d981aa7d98bb42268600c0109cdc6a968187bbddc8ba34553fa
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 9, 2026.
Transparency logSecurity for kingfisher-secret-scanner 1.6.3
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
All 7 files in this release have been uploaded with provenance metadata.
Attestation for kingfisher_secret_scanner-1.6.3.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 9, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.