Metraipolis › archive › dc_4fa36ee8761f

Authority Is Not a String: A Capability-Scoped Harness for Prompt-Injection-Resistant Coding Agents — arXiv:2609.08371

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.08371
License (as archived)
CC0-1.0
Archived
2026-10-16T14:09:11.778Z
SHA-256 of the archived text
8e541495ae2bd8b31ad28356c53ca37ef17ec5e7153dd294d3c2077e366a694d
Stable URL of this page
https://metraipolis.com/documento/dc_4fa36ee8761f

How to cite

Authority Is Not a String: A Capability-Scoped Harness for Prompt-Injection-Resistant Coding Agents — arXiv:2609.08371 — arXiv (en), https://arxiv.org/abs/2609.08371 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_4fa36ee8761f (Metraipolis archive dc_4fa36ee8761f, sha256 8e541495ae2bd8b31ad28356c53ca37ef17ec5e7153dd294d3c2077e366a694d)
BibTeX
@misc{dc4fa36ee8761f,
  title = {Authority Is Not a String: A Capability-Scoped Harness for Prompt-Injection-Resistant Coding Agents — arXiv:2609.08371},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.08371},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_4fa36ee8761f; sha256 8e541495ae2bd8b31ad28356c53ca37ef17ec5e7153dd294d3c2077e366a694d},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.08371. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Coding agents use system-level tools to read files, execute commands, and modify source code. Within the agent's sandbox, these tools often carry ambient authority: naming a resource is sufficient to act on it. Indirect prompt injection exploits this authority by placing instructions in repository files or tool output that cause the agent to perform actions the user did not request. We propose CapScope, a harness-level authorization mechanism that restricts tool use without requiring the model to identify malicious text. Before repository contents or tool output are read, CapScope derives a task-wide authority ceiling from trusted input. It then assigns each agent a separate set of typed capabilities, stored outside the model's context. Every tool call is checked against the capabilities of the agent that issued it. Permissions assigned to one sub-agent are therefore not automatically available to another. An injection may cause an agent to request an action, but the request is blocked unless that agent already has the required capability. We implement CapScope on the Pi coding agent and evaluate it in a repair workflow where an orchestrator delegates subtasks to separate sub-agents. The evaluation covers five Python tasks, five injection surfaces, four authorization conditions, and three trials per cell (300 runs). The injected effect executes in 33-47/75 runs under the ambient-authority and global-policy baselines, compared with 3/75 under CapScope. CapScope completes 68/75 repairs, while the baselines complete 68-72/75.

Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-09 in cs.SE).
Categorie arXiv: Software Engineering (cs.SE).
Pagina del lavoro: https://arxiv.org/abs/2609.08371 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-16T13:58:36.413Z dalla pagina https://arxiv.org/catchup/cs.SE/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).