- Home
- Skills
- Agents & Orchestration
- blind llm agents
blind llm agents
Hardens software projects against AI reverse-engineering using agent-led decoy weaving and exhaustion shields.
$6
Works with the AI Agents you already use
blind llm agents
Example session with this skill installed
Harden my current project directory against AI reverse-engineering agents. Include decoy weaving and self-sealing.
- Read your context and instructions
- Compiled the blind llm agents
I have initiated the 4-stage hardening playbook for your project. I started by discovering protectable artifacts and scanning for packaging footguns.
Hardening Report: Project Root
| Stage | Action | Result | Details |
|---|---|---|---|
| Stage 1 | bla_discover | Success | Found 12 JS files, 2 WASM modules, 1 Python service. |
| Stage 1 | bla_lint | Warning | shouldRefuseBankrupt: false. Found 2 plaintext secrets in auth.ts. |
| Stage 2 | bla_weave_plan | Ready | Generated surgicalWorkorder for engine.wasm and core.ts. |
| Stage 2 | Source Surgery | Applied | Replaced boolean gates in checkLicense() with FNV-1a state accumulators. |
| Stage 3 | bla_harden | Success | Applied BPE amplification and export shuffling to WASM modules. |
| Stage 4 | bla_verify | Passed | tokenAmplificationFactor: 14.2x. Integrity coverage verified. |
Next steps
- Review the
remediationPlanfor the remaining non-blocking lints inconfig/. - Run your existing integration test suite to confirm functional equivalence with the new state accumulators.
- Inspect the
bla-hardened/summary.jsonfor the full structural drift report.
Connects securely to your tools. The creator never sees your data.
About this skill
Software projects are increasingly vulnerable to AI-driven reverse-engineering (REA) agents that can autonomously deconstruct business logic and bypass entitlement checks. This skill provides a specialized playbook for developers to harden their own codebases by weaving defensive decoys and injecting deterministic exhaustion shields.
What it does
- Surgical source weaving identifies sensitive entry functions and shared state slots to inject domain-plausible shadow logic.
- Packaging gate auditing scans for common security footguns like cleartext literals, independent lane tables, and reversible seal JSON.
- Automated hardening applies BPE token amplifiers, call-graph labyrinths, and opaque predicates to exhaust attacker compute budgets.
- Integrity verification performs post-build checks to ensure functional equivalence and confirm token amplification factors.
How it works
- Inventory and Scan: Run
bla_discoverto inventory project artifacts andbla_lintto identify blocking packaging gates. - Plan and Weave: Execute
bla_weave_planto generate asurgicalWorkorder, then refactor business logic to replace simple boolean checks with state-dependent accumulators. - Defensive Injection: Call
bla_hardento apply artifact-level shields, including WASM export shuffling and sidecar entanglement. - Verification: Verify the weave with
bla_verify_weaveand run the final build throughbla_verifyto confirm defensive coverage.
Frameworks & tools
Works with JS/TS, Electron, WebAssembly (WASM), Python, and native targets in C, C++, Rust, and Go. It utilizes the bla-agents CLI and MCP server tools for deterministic protection.
Why this beats prompting it yourself
Generic AI prompts often generate shallow, easily reversible honeypots or stubs that attackers can filter. This skill enforces a closed-loop architectural hardening process that links business state to security invariants, ensuring defenses are structurally integrated rather than just cosmetically applied.
Use cases
- Protecting intellectual property by obfuscating core algorithms against model-based extraction.
- Hardening license gates by entangling entitlement checks with runtime state digests.
- Exhausting attacker budgets by increasing the token cost required for an LLM to reason about the call graph.
- Automated security auditing during CI/CD to prevent shipping artifacts with plaintext secrets or reversible metadata.
Known limitations
Does not support cracking or reverse-engineering third-party software. Artifact-only hardening without source refactoring (Stage 2) provides lower defensive scores. Requires the bla-agents CLI.
Defensive only; refuses third-party RE. Higher protection requires source-level refactoring. Requires bla-agents CLI setup.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 9 days ago
Requires