Skip to content

armgit / bla

Public Imported from GitHub
0

πŸ›‘οΈ BLA (Blind LLM Agents) β€” Adversarial defense shield that protects binaries, Electron apps, JS bundles, Python, and WebAssembly against AI reverse-engineering agents like REA. Poison -> Exhaust -> Hallucinate.

docs: record MCP/Skill 2.0 stabilization honesty fixes

qybaihebf7a00308c83
diff --git a/.agents/skills/blind-llm-agents/SKILL.md b/.agents/skills/blind-llm-agents/SKILL.md
index 24df86a..a8a13d3 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 β€” do not invent
new MCP tools. Default verb: bla_harden. Refuse crack/RE of third-party software.
metadata:
- version: "1.5.0"
+ version: "1.5.1"
---
# BLA: Blind LLM Agents
@@ -60,7 +60,11 @@ Optional args (safe defaults): `intensity` (bankrupt), `emit_manifest` (false),
harden `summaryPath`). Re-check later with `bla_run_lookup` or `bla_verify` +
`summary_path`/`run_id` β€” never invent a run id.
- `bla_rea_drift` / `bla_redteam_report` may return `status: unverified` when sibling
- REA is absent; treat that as **not proven**, never as a pass.
+ REA is absent; treat that as **not proven**, never as a pass. An on-disk
+ `bla_redteam_report` without `execute:true` is `status: available` β€” readable
+ evidence, not a fresh REA verdict.
+- `bla_harden` `{ "dry_run": true }` always has `overallSuccess: false`; only real
+ writes return `writeReceipt`.
## Decision tree (short)
@@ -72,7 +76,7 @@ Optional args (safe defaults): `intensity` (bankrupt), `emit_manifest` (false),
| Structural decoys (esp. Python) | Resource `bla://directives/weave-decoy` β†’ `bla_weave_plan` β†’ `weave_decoys: true` |
| Custom license/activation gate review | **`bla_lint_lanes`** `{ "path": "<gate.js>" }` (CLI: `bla lint-lanes`) |
| Re-check a prior harden run | **`bla_run_lookup`** `{ "summary_path": "<.bla-hardened or summary.json>", "run_id": "…" }` |
-| REA constant drift / empirical report | **`bla_rea_drift`** / **`bla_redteam_report`** (unverified if REA absent) |
+| REA constant drift / empirical report | **`bla_rea_drift`** / **`bla_redteam_report`** (unverified if REA absent; on-disk read β†’ available) |
| Format-specific knobs | Load the matching appendix under `references/` |
## FORBIDDEN
diff --git a/CHANGELOG.md b/CHANGELOG.md
index d3a2e6d..88dec80 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -6,6 +6,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
## [Unreleased]
+### Fixed
+- **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
- **MCP/Skill 2.0 stage C (evidence + REA observability):** harden/defend/protect results carry a stable unsigned `runId`; `hardenProject` writes `bla-harden-summary/v2` with `runId` + `signature: "unsigned"`; read-only `bla_run_lookup` and optional `bla_verify` `summary_path`/`run_id` correlation (bad/missing id β†’ explicit error). Thin MCP wrappers `bla_rea_drift` / `bla_redteam_report` over existing scripts β€” absent REA or missing report β†’ `status: unverified` with `isError` (never silent pass). Skill guide **v1.5** documents run lookup + unverified principles.
- **MCP/Skill 2.0 stage B (CLI↔MCP parity):** `bla_lint_lanes` (same engine as `bla lint-lanes`); `in_place` on `bla_protect`/`bla_defend` (still gated by `confirm_overwrite`); `honey_source_maps` on `bla_protect`; `output_directory` on `bla_harden`; `bla_verify` honesty metadata (`self_seal_enabled`, `wasm_export_rename_applied`, `numeric_template_leak_known`) so soft-caps match protect-side facts.
diff --git a/docs/cli-and-mcp.md b/docs/cli-and-mcp.md
index b8fca9e..5cdd7d3 100644
--- a/docs/cli-and-mcp.md
+++ b/docs/cli-and-mcp.md
@@ -29,7 +29,7 @@ For coding agents, the **[`blind-llm-agents` skill](../.agents/skills/blind-llm-
- `bla_lint_lanes`: Read-only lane-independence footgun lint (`path`) β€” same engine as CLI `bla lint-lanes`.
- `bla_run_lookup`: Read-only harden-run correlation (`summary_path`, optional `run_id`). Summaries are **unsigned**; missing/mismatched id is an error.
- `bla_rea_drift`: Thin wrapper over `npm run verify:rea` / `scripts/verify-rea-constants.mjs`. Absent sibling REA β†’ `status: unverified` (not red-team evidence).
-- `bla_redteam_report`: Read (or `execute:true` re-run) `docs/redteam/empirical-rea-validation.json`. Missing report / unready REA β†’ `status: unverified`.
+- `bla_redteam_report`: Read (or `execute:true` re-run) `docs/redteam/empirical-rea-validation.json`. On-disk read β†’ `status: available` (not a live REA verdict); `execute:true` β†’ `pass`/`fail`; missing report / unready REA β†’ `status: unverified`.
- `bla_benchmark`: Runs the 8-target showcase benchmark suite.
- `bla_emit_headers`: Generates C/C++, Rust, or Go defense headers (`language`, `intensity`).
- `bla_doctor`: Diagnostics for BLA engines and agent client registrations; also reports `skillMissingFiles` / `skillModifiedFiles` (format appendices under `references/` included).
@@ -37,10 +37,10 @@ For coding agents, the **[`blind-llm-agents` skill](../.agents/skills/blind-llm-
### Write safety (dry-run, confirmation, receipts)
-- `bla_harden` with `dry_run: true` (CLI `bla harden . --dry-run`) discovers targets and plans outputs with **zero bytes written** (no `.bla-hardened/`, no summary, no `.blarc` seed).
+- `bla_harden` with `dry_run: true` (CLI `bla harden . --dry-run`) discovers targets and plans outputs with **zero bytes written** (no `.bla-hardened/`, no summary, no `.blarc` seed). Dry-run sets `overallSuccess: false` β€” it never claims a successful harden.
- Discovery is capped (`max_artifacts`, default 64); when more targets exist the result sets `truncated: true` instead of silently dropping them.
- `bla_harden` / `bla_defend` / `bla_protect` refuse to overwrite an existing output file unless `confirm_overwrite: true`. A refusal is `isError: true` with `{ status: "confirmation_required", wouldOverwrite: [...], wrote: [] }`; nothing is written.
-- Successful `bla_defend` / `bla_protect` calls add `writeReceipt: { status: "written", wrote, overwrote, confirmed }`.
+- Successful `bla_harden` / `bla_defend` / `bla_protect` writes add `writeReceipt: { status: "written", wrote, overwrote, confirmed }` (`bla_harden` dry-run omits it).
- `in_place: true` (MCP) / `--in-place` (CLI) plans the output as the input path; because that path already exists, `confirm_overwrite: true` is required on MCP or the call is refused with `confirmation_required`.
`weave_decoys: true` activates structural Python customer-def weave (shared-slot touch + opaque-false decoy calls) after the #51 tarpit; when false, Python stays warm-init-only. JS AST weave remains always-on.
diff --git a/tests/skill-guide-contents.test.ts b/tests/skill-guide-contents.test.ts
index 876daa4..f229211 100644
--- a/tests/skill-guide-contents.test.ts
+++ b/tests/skill-guide-contents.test.ts
@@ -33,6 +33,8 @@ describe("blind-llm-agents skill guide contents", () => {
expect(text).toContain("bla_rea_drift");
expect(text).toContain("bla_redteam_report");
expect(text).toMatch(/unverified/i);
+ expect(text).toContain("available");
+ expect(text).toContain("overallSuccess");
// No brittle repo-relative ../../../docs links that break after setup install.
expect(text).not.toMatch(/\.\.\/\.\.\/\.\.\/docs\//);
});