dsh-output-custody-proof
Run the following command in DeepSeek Harness:
dsh plugin install dongsheng123132/dsh-output-custody-proof
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install dongsheng123132/dsh-output-custody-proof in the DeepSeek Harness terminal to install this plugin; the full source repository is at https://github.com/dongsheng123132/dsh-output-custody-proof
About this plugin
In the DSH framework, tool results pass through multiple stages — model-visible projection, spill storage, telemetry export — yet verifying whether each handoff is intact, whether truncation or mixing occurred, often lacks an offline, deterministic audit trail. dsh-output-custody-proof fills that gap: it never reads result bodies or opens spill artifacts, and instead reconciles SHA-256 digests, byte counts, projection classifications, and stage markers recorded in the receipt to validate the full chain from source to projection, from budget to omission.
Core checks cover byte budgets and exact omission accounting, spill and source identity consistency, surface mapping correctness, stage ordering, duplicate call and locator identity, and explicitly disclosed upstream incompleteness. The schema accepts only bounded identifiers, byte counts, classification labels, and digests — never raw tool values, prompts, credentials, or spill locators — eliminating sensitive data ingress at the interface level.
This plugin is best suited for developers and security auditors working with DSH 0.1.1-rc.1 or later, especially those handling mixed or image results, Code Mode sub-calls, and upstream truncation paths. If you need a deterministic verdict on a custody receipt without opening a single spill file or touching a result body, this plugin provides a lightweight, zero-side-effect verification tool.
Use Cases
- Offline-reconciling SHA-256 digests, byte budgets, and omission accounting in DSH custody receipts
- Auditing projection mapping and stage ordering in mixed or image result pipelines
- Issuing deterministic verdicts on upstream truncation disclosures without touching result bodies or spill artifacts
Best For
- Developers auditing the DSH tool-result lifecycle without access to result bodies
- Runtime engineers handling Code Mode sub-calls and upstream truncation
- Security audit teams performing deterministic compliance checks on custody receipts
Related Plugins
Traceable, searchable cross-session memory for AI agents that turns conversation knowledge into a typed knowledge graph and recalls relevant subgraphs instead of replaying full history, natively integrated with DeepSeek Harness.
Gives DSH AI cross-session long-term memory, to-do and skill management, plus multi-session orchestration, external AI delegation, and an infinite canvas that grows with you.
dsh-mnemon is a three-tier, pluggable, Agent-driven memory system for DeepSeek Harness, combining Runtime memory, Project Documents, and replaceable Memory Spaces with nine long-term providers.
An opinionated, zero-infrastructure file-based memory protocol: human-readable markdown files plus a bounded always-loaded index, curation discipline, and no database or embeddings.