What is a coding agent completion report?
A coding agent completion report is the agent's statement of what it says it changed and checked when it finishes a software task. In ET Verify, the machine-readable version of that report is the declaration in .et-verify/acceptance.json; ET Verify compares that declaration with repository state and configured-check evidence.
How it works
Completion report is the user-facing term. Declaration is the technical term for the machine-readable artifact that ET Verify evaluates.
- 01
Bind the claim to the work
The declaration binds its claims to a
baseCommit. ET Verify independently resolves the evaluated merge base and uses that trusted base context for configuration and evidence. - 02
Compare bounded facts
The declared exact-path scope and configured
TEST/BUILDclaims can be compared mechanically with repository state and trusted check execution where the evidence permits. - 03
Read the bounded result
ET Verify separates
AGENT CLAIMEDfromET ESTABLISHEDand reportsACCEPTED,REVIEW REQUIRED, orBLOCKED. The result is not a general correctness or security guarantee and does not establish everything the agent did.
Related answers
Install and documentation
Use the immutable public-alpha Action pin below. Start with the Quickstart for setup, then use Worked Examples to see the three results in context.
uses: empoweredtextlabs/et-verify@474a1d79599a714ccb2cac28a0b34c1444ad86c7
This is the Action coordinate. Setup also requires a GitHub workflow and your organization’s required-check configuration.