blitzstrike 2.4.62 → 2.4.64
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +34 -0
- package/dist/index.js +323 -148
- package/intelligence/variants.json +352 -0
- package/package.json +1 -1
package/CHANGELOG.md
CHANGED
|
@@ -5,6 +5,40 @@ All notable changes to Blitz Strike are documented here.
|
|
|
5
5
|
The format follows [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
|
|
6
6
|
and this project adheres to [Semantic Versioning](https://semver.org/).
|
|
7
7
|
|
|
8
|
+
## [2.4.64] — 2026-09-14
|
|
9
|
+
|
|
10
|
+
### Added — 47-pattern secret format taxonomy (severity + category)
|
|
11
|
+
|
|
12
|
+
- Expanded `SECRET_FORMATS` from ~12 to 47 known secret formats, each now carrying
|
|
13
|
+
a severity (critical/high/medium/low) + category (aws, gcp, github, stripe, slack,
|
|
14
|
+
email_svc, twilio, paas, firebase, jwt, bearer, basic_auth, private_key, generic,
|
|
15
|
+
ai_api, infra_api, package_registry, saas_api, observability, tunneling, bot_token).
|
|
16
|
+
- New formats: AWS session tokens, GCP service accounts, GitHub OAuth/server-to-server,
|
|
17
|
+
Stripe live/test/publishable, Slack webhooks, Heroku, Firebase, bearer/basic-auth,
|
|
18
|
+
RSA/EC/OpenSSH private keys, Anthropic/OpenAI/HuggingFace, Cloudflare/DigitalOcean,
|
|
19
|
+
npm/PyPI/Docker, Atlassian/Linear, NewRelic/Datadog/Sentry, ngrok, Discord/Telegram.
|
|
20
|
+
- `detectHardcodedSecret` Tier 2 now reports the format's own severity + category,
|
|
21
|
+
so a hardcoded secret hit is classified (e.g. Stripe live = critical, Stripe test
|
|
22
|
+
= low) instead of uniformly "high". Redaction patterns expanded to match.
|
|
23
|
+
|
|
24
|
+
## [2.4.63] — 2026-09-14
|
|
25
|
+
|
|
26
|
+
### Added — Variant corpus + detector coverage sweep (data-driven scaling)
|
|
27
|
+
|
|
28
|
+
- New `intelligence/variants.json` — a corpus of 65 concrete real-world code
|
|
29
|
+
shapes (3-4 per class, multi-language: PHP/JS/Python/Java/Node) that each MUST
|
|
30
|
+
fire its class's detector. This is how the empirical grounding scales: more
|
|
31
|
+
shapes → more coverage, tested reproducibly.
|
|
32
|
+
- New `scan_variants` MCP tool + `blitzstrike verify` now also sweeps the variant
|
|
33
|
+
corpus and reports per-class coverage, surfacing false negatives (a detector
|
|
34
|
+
missing a real shape) as a hard failure.
|
|
35
|
+
- Detector coverage raised from 54% to 100% across the corpus, fixing real gaps:
|
|
36
|
+
multi-language sinks (pickle/yaml/readObject, jinja2/twig, axios/requests,
|
|
37
|
+
DocumentBuilderFactory), `$.extend`/`$twig->render` word-boundary bugs, JS `+`
|
|
38
|
+
concatenation in SQLi, `setcookie` CRLF, `mysqli_query`/`whereRaw` SQLi sinks,
|
|
39
|
+
dynamic field binding + `Model.create` mass-assignment, and a double-paren bug
|
|
40
|
+
in the upload sink regex. The corpus is the proof: every shape now fires.
|
|
41
|
+
|
|
8
42
|
## [2.4.62] — 2026-09-14
|
|
9
43
|
|
|
10
44
|
### Added — Deep empirical corpus (per-class bypass + crown jewels)
|