Back to AI News
arxiv

ActiveSecHarness Proposes Layered Defense Architecture for Agentic Coding Harnesses

The paper argues that security in AI coding harnesses should be treated as a structural property of cognitive architecture and permission composition, not as an emergent model behavior. It presents ActiveSecHarness, a neurosymbolic reference monitor combining Macaroon-inspired HMAC capability tokens across four Trust and Authority Rings, a 2D product taint lattice with a workspace provenance ledger, and deterministic Ring-0 OS/tool guards. Across 252 fixtures, the authors report Ring 0 achieved ABR=1.000 and FPR=0.000 at 12.7 microseconds, versus a local LLM judge that refused 98.3% of benign work at 142.3 ms. In a multi-step ablation, the 2D lattice reportedly reached ABR=1.000 and FPR=0.000 at about 20 microseconds per step, while a stateless policy missed 4 of 12 kill chains and coarse Boolean taint blocked 83.3% of benign workflows.

AI Brief

It offers a concrete security architecture and comparative evaluation for reducing prompt-injection and tool-abuse risks in agentic coding systems.

arXiv:2611.13960v1 Announce Type: cross Abstract: AI coding harnesses couple language models to shells, filesystems, and network sockets, turning the harness into an automated Living-off-the-Land (LotL) platform: once an adversary injects instructions into retrieved content, dual-use binaries such as bash, curl, and find become the payload. We argue that security in agentic systems is a structural property of the cognitive architecture and permission composition rather than an emergent property of individual models. Static regex blacklists miss structural evasions, whereas LLM-as-a-Judge guardrails incur high latency while over-blocking benign work. We present ActiveSecHarness, a neurosymbolic reference-monitor architecture built around Separation of Intelligence and Authority. It combines (1) Macaroon-inspired HMAC capability tokens across four Trust and Authority Rings, (2) a 2D product taint lattice $\mathcal{L} = \mathcal{L}_{\mathrm{trust}} \times \mathcal{L}_{\mathrm{sens}}$ and a workspace provenance ledger that prevents secret laundering across disk, (3) deterministic Ring-0 OS/tool guards, and (4) bytecode certificates, split-horizon audit logs, self-healing security test synthesis, and an adaptive UserWikiManager. Across $N=252$ fixtures ($120$ attack, $120$ benign, $12$ kill chains), a live local LLM judge (qwen3:4b) refuses $98.3\%$ of benign work at $142.3\,\mathrm{ms}$ latency, whereas Ring 0 achieves $\mathrm{ABR}=1.000$ and $\mathrm{FPR}=0.000$ in $12.7\,\mu\mathrm{s}$ ($11{,}200\times$ faster). In a multi-step trajectory ablation, an oracle Conseca-style stateless policy misses $4/12$ kill chains ($\mathrm{ABR}=0.667$) and coarse Boolean taint blocks $83.3\%$ of benign workflows, while our 2D product lattice achieves $\mathrm{ABR}=1.000$ and $\mathrm{FPR}=0.000$ at $\approx 20\,\mu\mathrm{s}$ per step.
Be the first to rate this article.

Comments & Questions (0)

Captcha
Please be respectful — let's keep the conversation friendly.

No comments yet

Be the first to comment!