chore(release): bla-agents v2.0.0

Bump package/identity/lock to 2.0.0; CHANGELOG + STATE + Skill v2.0.0
document packaging-gates, MCP write-safety, and R1–R12 refuse/recipe/honesty
vs 1.x. Release notes in docs/releases/v2.0.0.md.
This commit is contained in:
qybaihe
2026-11-03 11:56:33 +08:00
parent 9005878eb9
commit c541f0be09
8 changed files with 65 additions and 10 deletions
+3 -1
View File
@@ -7,7 +7,7 @@ description: >
Agents). Prefer this skill as the guide; call existing MCP tools / CLI — do not invent
new MCP tools. Default verb: bla_harden. Refuse crack/RE of third-party software.
metadata:
version: "1.5.23"
version: "2.0.0"
---
# BLA: Blind LLM Agents
@@ -24,6 +24,8 @@ model misjudgment on proprietary logic, or **one-click / 一键加固** 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, R1–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 — no fake refuse lints.
**Principles (red-blue R1–R12 → MCP/Skill 2.0):**
### A. Bankrupt refuse gates (engine + MCP — call `bla_lint` first)
+10 -1
View File
@@ -6,8 +6,16 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
## [Unreleased]
## [2.0.0] — 2026-11-03
Major MCP + Skill 2.0 release. Red/Blue R1–R12 packaging gates that held under adversarial
dogfood are now product defaults: unified `bla lint` / `bla_lint`, bankrupt refuse lints,
WASM export hygiene, agent write-safety, and honest Skill recipes for crypto that cannot
be a packaging lint. Breaking only in the sense of new bankrupt refuses / ABI defaults —
CLI/MCP verbs remain `bla_harden` / `bla_defend` / `bla_protect`.
### Added
- **Unified packaging gates (`bla lint` / MCP `bla_lint`):** one project scan over every R1–R7 bankrupt refuse lint plus lane/panel discovery (`bla-packaging-gates/v1`). `hardenProject` shares `evaluateHardenProjectPackagingBlocks`. Skill guide **v1.5.23** layers refuse gates / product recipes / honesty residual from red-blue R1–R12. WASM protect honesty surfaces `wasmExportOrderShuffled` / `wasmExportDecoysInjected`. MCP parity includes `allow_wasm_seal_xor_known_plaintext` on harden/defend/protect.
- **Unified packaging gates (`bla lint` / MCP `bla_lint`):** one project scan over every R1–R7 bankrupt refuse lint plus lane/panel discovery (`bla-packaging-gates/v1`). `hardenProject` shares `evaluateHardenProjectPackagingBlocks`. Skill guide **v2.0.0** layers refuse gates / product recipes / honesty residual from red-blue R1–R12. WASM protect honesty surfaces `wasmExportOrderShuffled` / `wasmExportDecoysInjected`. MCP parity includes `allow_wasm_seal_xor_known_plaintext` on harden/defend/protect.
- **Red/Blue round-4 · Writeback — JS AES unwrap lint + panel offline-oracle gate:** CLI `bla lint-js-aes-unwrap` / MCP `bla_lint_js_aes_unwrap` detect hex/base64 wrap-key material + `createDecipheriv`/`createCipheriv` + fixed sealed hdr (e.g. `d17ea004`). Bankrupt harden/protect **refuse** unless `allowJsAesUnwrap` / `--allow-js-aes-unwrap` / `allow_js_aes_unwrap`. Gate only — **not** white-box AES. `bla_lint_panel_derive` / CLI also flag stable `deriveLocal` (maximum/bankrupt rewrite to `_blaPdLocal_*`) and `deriveLocal` + plaintext `panelTokenSha256` offline-oracle packaging (same-file or sibling `seal.json`); bankrupt **refuses** the oracle combo unless `allowPanelOfflineOracle` / `--allow-panel-offline-oracle` / `allow_panel_offline_oracle`. Skill **v1.5.7**. Self-seal sibling SHA-256 hard-fail remains from `c2ec868`.
- **Red/Blue round-3 · Writeback residual — WASM param lint + panel rewrite + self_seal recommend:** CLI `bla lint-wasm-params` / MCP `bla_lint_wasm_params` detect clear `R3PARM`/`R3CIPH`/`R2SEAL`-class ASCII carriers with adjacent binary blobs; bankrupt harden/protect **refuse** unless `allowWasmClearParamCarriers` / `--allow-wasm-clear-params` / `allow_wasm_clear_params`. CLI `bla lint-panel-derive` / MCP `bla_lint_panel_derive` flag stable `__vaultPanelDerive_*` (and classic FNV primes as warning); maximum/bankrupt protect **rewrites** to seed-stable `_blaPd_*` unless `allowStablePanelDeriveAbi` / `--allow-stable-panel-derive` / `allow_stable_panel_derive`. Bankrupt harden sets `selfSealRecommended` for multi-artifact JS/WASM/HTML trees without auto-enabling single-file; `--self-seal` on harden auto-binds sibling paths. Audit surfaces the same WASM/panel findings. Skill **v1.5.6**. Full WASM data-section encryption remains a follow-up.
- **Red/Blue round-3 · Red writeback (skill):** one-way SHA-256 seals **close** the round-2 ARX-invert break; skill **v1.5.5** + format appendices document residual packaging — WASM `R3PARM`/`R3CIPH`-class clear param/cipher carriers, stable `__vaultPanelDerive_*` / FNV literals, and prefer `self_seal` for multi-artifact entitlement trees. No emergency P0.
@@ -19,6 +27,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
- **Red/Blue round-1 · Writeback P0 — lane-independence is error + harden refuse:** `bla lint-lanes` / `bla_lint_lanes` findings are severity **error**; protect/defend/harden refuse JS/TS that looks lane-independent (including per-byte WASM mix/seal oracles) unless `allowLaneIndependentGates` / `--allow-lane-independent` / `allow_lane_independent`. Skill v1.5.2 principles: no silent stable WASM ABI; no JS independent lane tables; prefer whole-key mixing.
### Fixed
- **CI hydrate-dump lint fixtures:** optional `/workspace/bla-redblue` dogfood paths no longer fail clones; in-repo synthetic R5/R6 shapes cover refuse vs unlock-in-WASM.
- **MCP/Skill 2.0 stabilization:** `bla_harden` dry-run no longer sets `overallSuccess: true`; CLI formatter labels dry-run as `DRY-RUN (nothing written)` instead of `SUCCESS`. Successful `bla_harden` writes now return `writeReceipt` (matching defend/protect and the stage A contract). `bla_redteam_report` on-disk reads use `status: available` (not `pass`) so agents cannot mistake a static fixture for a live REA verdict; harness stdout/stderr are clipped for MCP stdio safety. Skill guide **v1.5.1** documents available vs unverified.
### Added
+2 -2
View File
@@ -1,6 +1,6 @@
# BLA state of the project
**Version:** 1.6.4 (`package.json` ↔ `src/identity.ts`, verified by build; see `CHANGELOG.md`)
**Version:** 2.0.0 (`package.json` ↔ `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.
@@ -89,7 +89,7 @@ Outside the repo, `npm pack` → install tarball → isolated `HOME`:
## Red-blue packaging gates (R1–R12 → 2.0)
Unified `bla lint` / MCP `bla_lint` (`bla-packaging-gates/v1`) aggregates every bankrupt refuse lint from red-blue rounds 1–7; R8–R12 SHA-256 KDF / commitment-outside-XOR / arity+export hygiene live as Skill recipes + WASM protect defaults. Honesty residual: full WASM RE + online/SMT — **not** a new refuse lint. Skill **v1.5.23**. Draft PR #77 — do not merge/ship 2.0 without maintainer approval.
Unified `bla lint` / MCP `bla_lint` (`bla-packaging-gates/v1`) aggregates every bankrupt refuse lint from red-blue rounds 1–7; R8–R12 SHA-256 KDF / commitment-outside-XOR / arity+export hygiene live as Skill recipes + WASM protect defaults. Honesty residual: full WASM RE + online/SMT — **not** a new refuse lint. Skill **v2.0.0**. Shipped as **bla-agents 2.0.0** (PR #77).
## Nightly roadmap remaining
+44
View File
@@ -0,0 +1,44 @@
# bla-agents v2.0.0 — MCP + Skill 2.0
**Date:** 2026-11-03 (Asia/Singapore)
## Headline
Red/Blue R1–R12 packaging that held under adversarial dogfood is now product:
- Unified **`bla lint` / MCP `bla_lint`** (`bla-packaging-gates/v1`) + shared harden blocks
- Bankrupt refuse lints for lanes, seals, WASM carriers, JS AES unwrap, panel oracles, hydrate dumps, XOR-mask sidecars, seal.revision XOR-KP
- WASM maximum/bankrupt defaults: export rename, arity-2-preferring decoys, independent export-order shuffle
- MCP/CLI write-safety (`discover`, `dry_run`, `confirm_overwrite`, `writeReceipt`) + run evidence
- Skill **v2.0.0** layers refuse gates / product recipes / honesty residual (full WASM RE + online/SMT — no fake lint)
## Relative to 1.x
| Area | 1.x | 2.0.0 |
|------|-----|-------|
| Packaging scan | Per-lint CLIs | Unified `bla lint` / `bla_lint` |
| Bankrupt footguns | Partial (grew across 1.5–1.6) | Full R1–R7 refuse set + WASM hygiene |
| Crypto recipes (KDF, commitment, pad3) | Ad-hoc skill notes | Explicit Skill B-layer (not refuse lints) |
| Agent overwrite / dry-run | Incomplete | Stage A–C contract |
## Gates
- `npm run build` — required green
- `npm test` — ~467 pass / 2 skip
- Tag: `bla-agents-2.0.0` (and `v2.0.0` alias)
## Publish checklist
```bash
gh auth status
git checkout main && git pull
git tag -a bla-agents-2.0.0 -m "bla-agents v2.0.0"
git tag -a v2.0.0 -m "bla-agents v2.0.0"
git push origin bla-agents-2.0.0 v2.0.0
gh release create bla-agents-2.0.0 --title "bla-agents v2.0.0" --notes-file docs/releases/v2.0.0.md
# npm publish only when npm credentials are configured
```
## Red line
Unsigned APKs only. Defensive tooling — refuse crack/RE of third-party software.
+2 -2
View File
@@ -1,12 +1,12 @@
{
"name": "bla-agents",
"version": "1.6.4",
"version": "2.0.0",
"lockfileVersion": 3,
"requires": true,
"packages": {
"": {
"name": "bla-agents",
"version": "1.6.4",
"version": "2.0.0",
"license": "MIT",
"dependencies": {
"@babel/generator": "^7.26.0",
+1 -1
View File
@@ -1,6 +1,6 @@
{
"name": "bla-agents",
"version": "1.6.4",
"version": "2.0.0",
"description": "BLA (Blind LLM Agents) \u2014 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",
+1 -1
View File
@@ -5,7 +5,7 @@ export const BLA_IDENTITY = Object.freeze({
packageAlias: "bla-shield",
cliBinary: "bla",
cliAlias: "bla-agents",
version: "1.6.4",
version: "2.0.0",
slogan: "Poison -> Exhaust -> Hallucinate. Make Reverse Engineering Bankrupt Again.",
counterpart: {
name: "REA",
+2 -2
View File
@@ -11,9 +11,9 @@ const skillDir = path.resolve(
const skillPath = path.join(skillDir, "SKILL.md");
describe("blind-llm-agents skill guide contents", () => {
it("ships v1.5+ short decision guide with write-safety, lint-lanes, and run evidence", () => {
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+"/);
expect(text).toMatch(/metadata:\s*\n\s*version:\s*"(?:1\.5\.\d+|2\.0\.\d+)"/);
expect(text).toContain("bla_harden");
expect(text).toContain("npx bla-agents@latest harden .");
expect(text).toContain("weave_decoys");