AI Agent Hub
Back to plugins
⚙️

dsh-surface-contract-proof

Workflow Updated 2026.08.20

Run the following command in DeepSeek Harness:

dsh plugin install dongsheng123132/dsh-surface-contract-proof

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install in the DeepSeek Harness terminal to add this plugin; the full source is at https://github.com/dongsheng123132/dsh-surface-contract-proof, and the inspect and verify tools become available after a session restart.

About this plugin

After upgrading the DSH ToolRuntime, MCP, or CLI pipeline, the key question is whether recorded request/response envelopes still preserve the same semantic contract. dsh-surface-contract-proof offline-verifies SHA-256-pinned JSON fixtures across all three machine surfaces, checking schema versions, success/conflict bits, exit-code mapping, timeout and out-of-order behavior, and result digests without executing targets, calling the network, or replaying side effects.

Safety guarantees: all three surface types must be present; fixtures must bind target revisions and byte hashes; missing, stale, or drifted fixtures fail closed. Reports expose only hashes, identities, statuses, and differing field names-never request arguments, response bodies, or credentials. Paths are restricted to workspace-relative regular files with atomic read-back verification.

Ideal for DSH developers and platform engineers performing upgrade regressions or contract audits. Unlike generic Pact or OpenAPI tools, it is purpose-built for the three DSH machine surfaces and operates fully offline.

Use Cases

  • Offline-compare recorded envelopes after upgrading ToolRuntime, MCP, or CLI to confirm the semantic contract has not drifted
  • Fail-closed validation of schema versions, exit-code mapping, and conflict semantics with SHA-256-pinned fixtures before release
  • Audit whether success, conflict, and timeout behaviors remain identical across all three DSH surfaces between revisions

Best For

  • Platform engineers maintaining DSH ToolRuntime or MCP services
  • Release managers who need repeatable offline regression evidence
  • Developers performing contract audits and upgrade decisions in the DSH ecosystem