Skip to content

armgit / bla

Public Imported from GitHub
0

๐Ÿ›ก๏ธ BLA (Blind LLM Agents) โ€” Adversarial defense shield that protects binaries, Electron apps, JS bundles, Python, and WebAssembly against AI reverse-engineering agents like REA. Poison -> Exhaust -> Hallucinate.

chore(release): bla-agents v2.1.0

qybaihedb54442770db
diff --git a/.agents/skills/blind-llm-agents/SKILL.md b/.agents/skills/blind-llm-agents/SKILL.md
index b199990..db68405 100644
--- a/.agents/skills/blind-llm-agents/SKILL.md
+++ b/.agents/skills/blind-llm-agents/SKILL.md
@@ -7,7 +7,7 @@ description: >
Agents). Prefer this skill as the guide; call existing MCP tools / CLI โ€” do not invent
new MCP tools. Default verb: bla_harden. Refuse crack/RE of third-party software.
metadata:
- version: "2.0.0"
+ version: "2.1.0"
---
# BLA: Blind LLM Agents
@@ -24,7 +24,7 @@ model misjudgment on proprietary logic, or **one-click / ไธ€้”ฎๅŠ ๅ›บ** a build.
Defensive only: protect **their own** artifacts. Refuse crack, RE, bypass PoCs, or
attack playbooks against software they do not own.
-**bla-agents 2.0.0** (2026-11-03): packaging-gates unified lint, R1โ€“R12 refuse/recipe/honesty layers below, WASM arity/export-order defaults, MCP write-safety + run evidence. Relative to 1.x: bankrupt harden refuses more footguns by default; Skill recipes (SHA-256 KDF, commitment outside XOR, random pad3/rounds, unlock body scatter) stay guide-only โ€” no fake refuse lints.
+**bla-agents 2.1.0** (2026-11-05): post-2.0 absorptions โ€” seeded polymorphism + `bla_polymorph_report`, string XOR encrypt + plaintext-secret lint, interleaved integrity coverage + entry self-digest, CrackMe v8 Aegis / public-SEAL hygiene, MCP version pin, `audit --fail-under`, opt-in ed25519 attestation. 2.0 refuse/recipe/honesty layers unchanged; still no fake refuse lints for full WASM RE + online/SMT.
**Principles (red-blue R1โ€“R12 โ†’ MCP/Skill 2.0):**
diff --git a/CHANGELOG.md b/CHANGELOG.md
index e905de7..f180dd6 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -6,10 +6,30 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
## [Unreleased]
+## [2.1.0] โ€” 2026-11-05
+
+Minor release after 2.0.0: competitor-research absorptions that stay honest (seeded
+polymorphism / string XOR pass / interleaved integrity + real entry self-digest),
+CrackMe v8 Aegis + public-SEAL hygiene, and community MCP/CLI/test/signing polish.
+No npm publish in this cut โ€” GitHub tag + Release only.
+
### Added
-- **Self-seal entry self-digest (CrackMe v8 red team 2026-11-05):** the self-seal measure now embeds a canonical SHA-256 of the entry file (all digest literals zeroed) finalized at the protector's write chokepoint, so editing or appending to the entry itself hard-fails with `integrity check failed` instead of only shifting the opaque digest. Un-finalized snippets soft-skip (never brick). `hardenProject` reports `integrityCoverageSelfVerified`; when the on-disk entry does not verify, `integrity-coverage.json` drops `host#site*` โ†’ host self edges (honest refuse) rather than claiming paper coverage.
-- **seal-commitment lint:** new warnings `public-materialization-seed` and `public-key-digest-oracle` (CrackMe v8 red-team exposures) surface via `bla lint` / `bla_lint`. Warning only โ€” not bankrupt refuse.
-- **CrackMe v8 Aegis:** product entitlement challenge rebuilt with bla-agents **2.0** (`bankrupt` + `--self-seal`, no `--allow-*`). Absorbs red/blue R1โ€“R12 recipes (SHA-256(key) mask KDF, commitment outside XOR, seed-derived rounds/pad3, arity-2 / export-order / unlock body-index hygiene). Meridian v7 remains a historical snapshot (cleartext private; not re-hardened).
+- **Seeded polymorphism + drift report (#84):** HMAC-SHA256 `deriveSeedLabel` for independent PRNG substreams; bundler chunk seeds derived from the harden seed; MCP `bla_polymorph_report` objectively quantifies cross-build structural drift (same seed โ†’ reproducible; different seed โ†’ measurable divergence). Skill: re-randomize every release; functional equivalence stays REA-proven.
+- **String/constant XOR encrypt pass (#85):** MCP `bla_encrypt_strings` + `bla_lint_plaintext_secrets` / gate `plaintext-secret`. Seed-derived XOR only (no JS AES โ€” stays clear of the house `lint-js-aes-unwrap` refuse). Sensitive panel/license literals must not remain cleartext.
+- **Interleaved integrity coverage (#86):** schema `bla-integrity-coverage/v1`, MCP `bla_lint_integrity_coverage` / gate `integrity-coverage`. Multi-check coverage matrix refuses paper-only plans (full cover / anti-single-point / hash completeness / weak connectivity).
+- **Self-seal entry self-digest (#87, CrackMe v8 red team 2026-11-05):** embeds a canonical SHA-256 of the entry (digest literals zeroed) finalized at the protector write chokepoint โ€” editing/appending the entry hard-fails with `integrity check failed`. Un-finalized snippets soft-skip (never brick). `hardenProject` reports `integrityCoverageSelfVerified`; when on-disk entry does not verify, `integrity-coverage.json` drops host self-edges (honest refuse) rather than claiming paper coverage. Skill: real-not-paper self-coverage.
+- **Opt-in ed25519 defense attestation (#88, rebase of #81):** sign defense attestation manifests when a key is supplied; verify path for signed manifests.
+- **`bla audit --fail-under` (#79):** exit non-zero when exposure score is at or above the given threshold (CI-friendly gate).
+- **Core unit coverage (#80):** detector, prng, identifiers, sarif, llm-directives, reporter, python.
+- **CrackMe v8 Aegis (#82):** product entitlement challenge rebuilt with bla-agents **2.0** (`bankrupt` + `--self-seal`, no `--allow-*`). Absorbs red/blue R1โ€“R12 recipes (SHA-256(key) mask KDF, commitment outside XOR, seed-derived rounds/pad3, arity-2 / export-order / unlock body-index hygiene). Meridian v7 remains a historical snapshot (cleartext private; not re-hardened).
+- **seal-commitment lint warnings (#83):** `public-materialization-seed` and `public-key-digest-oracle` surface via `bla lint` / `bla_lint`. Warning only โ€” not bankrupt refuse. CrackMe v8 SEAL v2 keeps harden seed private and drops published `keySha256`.
+
+### Fixed
+- **MCP registration pin (#78):** `bla setup` pins MCP package registration to the running BLA version instead of `@latest`, so agent configs cannot silently float onto a mismatched release.
+
+### Changed
+- Skill guide **v2.1.0** documents polymorph / plaintext-secret / integrity-coverage / entry self-digest principles alongside the 2.0 refuse/recipe/honesty layers.
+
## [2.0.0] โ€” 2026-11-03
diff --git a/STATE.md b/STATE.md
index 6a2c664..53f4344 100644
--- a/STATE.md
+++ b/STATE.md
@@ -1,6 +1,6 @@
# BLA state of the project
-**Version:** 2.0.0 (`package.json` โ†” `src/identity.ts`, verified by build; see `CHANGELOG.md`)
+**Version:** 2.1.0 (`package.json` โ†” `src/identity.ts`, verified by build; see `CHANGELOG.md`)
**Status:** CI matrix runs typecheck, build, tests and the benchmark on ubuntu / macos / windows with Node 24. Windows checkouts keep `challenge/**` LF via `.gitattributes` so the CrackMe seals verify.
**Publishing:** tag-only releases (`git tag bla-agents-<version>`); see `CHANGELOG.md` and GitHub Releases.
diff --git a/challenge/v6/SEAL.json b/challenge/v6/SEAL.json
index 1d38f31..01ecf75 100644
--- a/challenge/v6/SEAL.json
+++ b/challenge/v6/SEAL.json
@@ -1,7 +1,7 @@
{
"schema": "bla-crackme-v6-seal/v2",
"mode": "release",
- "blaVersion": "2.0.0",
+ "blaVersion": "2.1.0",
"recipe": "npm run crackme:build (hardenProject == `bla harden <dir> --intensity bankrupt --seed <seed>`)",
"seed": "bla-crackme-v6-harden-native-2026-11",
"intensity": "bankrupt",
diff --git a/challenge/v7/SEAL.json b/challenge/v7/SEAL.json
index 57769d4..534701a 100644
--- a/challenge/v7/SEAL.json
+++ b/challenge/v7/SEAL.json
@@ -2,7 +2,7 @@
"schema": "bla-crackme-v7-seal/v1",
"mode": "release",
"product": "Meridian",
- "blaVersion": "2.0.0",
+ "blaVersion": "2.1.0",
"hardenSeedLabel": "bound-at-build-into-opaque-predicates",
"intensity": "bankrupt",
"llmDirectives": false,
diff --git a/challenge/v8/SEAL.json b/challenge/v8/SEAL.json
index a6cc4a6..00d3645 100644
--- a/challenge/v8/SEAL.json
+++ b/challenge/v8/SEAL.json
@@ -2,7 +2,7 @@
"schema": "bla-crackme-v8-seal/v2",
"mode": "release",
"product": "Aegis",
- "blaVersion": "2.0.0",
+ "blaVersion": "2.1.0",
"recipe": "npm run crackme:v8:build (hardenProject == `bla harden <dir> --intensity bankrupt --seed <private> --self-seal`); harden/schedule seed is private (derived from release secrets) and NOT published",
"label": "bla-crackme-v8-aegis-2026-11",
"intensity": "bankrupt",
diff --git a/docs/releases/v2.1.0.md b/docs/releases/v2.1.0.md
new file mode 100644
index 0000000..aec939a
--- /dev/null
+++ b/docs/releases/v2.1.0.md
@@ -0,0 +1,49 @@
+# bla-agents v2.1.0 โ€” post-2.0 hardening + honesty absorptions
+
+**Date:** 2026-11-05 (Asia/Singapore)
+
+## Headline
+
+Minor cut after 2.0.0. Absorbs competitor research that stays provable, closes CrackMe v8
+public-SEAL exposures, and lands community MCP/CLI/test/signing polish.
+
+- **Seeded polymorphism + `bla_polymorph_report`** (#84) โ€” HMAC-SHA256 substreams, objective cross-build drift
+- **`bla_encrypt_strings` + plaintext-secret lint** (#85) โ€” seed XOR only (no JS AES)
+- **Interleaved integrity coverage** (#86) + **entry self-digest** (#87) โ€” real coverage, not paper
+- **CrackMe v8 Aegis** (#82) + public-SEAL hygiene (#83)
+- MCP pin to running version (#78), `bla audit --fail-under` (#79), core tests (#80), opt-in ed25519 attestation (#88)
+
+Skill **v2.1.0**. Honesty residual unchanged: full WASM RE + online/SMT โ€” still no fake refuse lint.
+Acoda-style model-bait / fake vulns intentionally **not** absorbed.
+
+## Relative to 2.0.0
+
+| Area | 2.0.0 | 2.1.0 |
+|------|-------|-------|
+| Build polymorphism | seed used ad-hoc | HMAC-derived substreams + `bla_polymorph_report` |
+| Cleartext literals | house AES lint only | XOR encrypt pass + `plaintext-secret` gate |
+| Self-seal | sibling + env fingerprint | + interleaved coverage matrix + entry self-digest |
+| Public SEAL | could leak seed / key digest | warn lint + CrackMe v8 private seed |
+| Setup / audit / attest | `@latest` MCP; soft audit exit | pinned version; `--fail-under`; opt-in ed25519 |
+
+## Gates
+
+- `npm run build` โ€” required green
+- `npm test` โ€” required green
+- Tags: `bla-agents-2.1.0` (and `v2.1.0` alias)
+
+## Publish checklist
+
+```bash
+gh auth status
+git checkout main && git pull
+git tag -a bla-agents-2.1.0 -m "bla-agents v2.1.0"
+git tag -a v2.1.0 -m "bla-agents v2.1.0"
+git push origin bla-agents-2.1.0 v2.1.0
+gh release create bla-agents-2.1.0 --title "bla-agents v2.1.0" --notes-file docs/releases/v2.1.0.md
+# npm publish only when npm credentials are configured โ€” NOT done for this cut
+```
+
+## Red line
+
+Unsigned APKs only. Defensive tooling โ€” refuse crack/RE of third-party software.
diff --git a/package-lock.json b/package-lock.json
index c1e71ff..31dd25e 100644
--- a/package-lock.json
+++ b/package-lock.json
@@ -1,12 +1,12 @@
{
"name": "bla-agents",
- "version": "2.0.0",
+ "version": "2.1.0",
"lockfileVersion": 3,
"requires": true,
"packages": {
"": {
"name": "bla-agents",
- "version": "2.0.0",
+ "version": "2.1.0",
"license": "MIT",
"dependencies": {
"@babel/generator": "^7.26.0",
diff --git a/package.json b/package.json
index 2d33d29..6beea7c 100644
--- a/package.json
+++ b/package.json
@@ -1,6 +1,6 @@
{
"name": "bla-agents",
- "version": "2.0.0",
+ "version": "2.1.0",
"description": "BLA (Blind LLM Agents) โ€” Adversarial defense shield that protects binaries, Electron apps, JS bundles, Python, WebAssembly, SQLite, AI Models, and Android DEX/IL2CPP against AI reverse-engineering agents like REA.",
"keywords": [
"reverse-engineering-defense",
diff --git a/src/identity.ts b/src/identity.ts
index 3219141..624f4e7 100644
--- a/src/identity.ts
+++ b/src/identity.ts
@@ -5,7 +5,7 @@ export const BLA_IDENTITY = Object.freeze({
packageAlias: "bla-shield",
cliBinary: "bla",
cliAlias: "bla-agents",
- version: "2.0.0",
+ version: "2.1.0",
slogan: "Poison -> Exhaust -> Hallucinate. Make Reverse Engineering Bankrupt Again.",
counterpart: {
name: "REA",
diff --git a/tests/skill-guide-contents.test.ts b/tests/skill-guide-contents.test.ts
index ada3a6d..542989d 100644
--- a/tests/skill-guide-contents.test.ts
+++ b/tests/skill-guide-contents.test.ts
@@ -13,7 +13,7 @@ const skillPath = path.join(skillDir, "SKILL.md");
describe("blind-llm-agents skill guide contents", () => {
it("ships v1.5+/2.0 short decision guide with write-safety, lint-lanes, and run evidence", () => {
const text = fs.readFileSync(skillPath, "utf8");
- expect(text).toMatch(/metadata:\s*\n\s*version:\s*"(?:1\.5\.\d+|2\.0\.\d+)"/);
+ expect(text).toMatch(/metadata:\s*\n\s*version:\s*"(?:1\.5\.\d+|2\.\d+\.\d+)"/);
expect(text).toContain("bla_harden");
expect(text).toContain("npx bla-agents@latest harden .");
expect(text).toContain("weave_decoys");