Repository navigation
Evidence as execution authority rather than decision evidence? #4159
Derrick Surratt (derricksurrattFFA)
started this conversation in
Ideas
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
I've been studying the distinction between evidence about a governance decision and evidence that causally determines whether an autonomous actor has authority to execute. AGT already appears to address policy evidence, effective capabilities, action-bound approval, trust state, replay, and tamper-evident audit.
I'm interested in a slightly different architecture:
Evidence state → evidence sufficiency → governance constraint → effective authority/capability → execution eligibility → action → proof/outcome → updated evidence state
In this model, evidence is not merely attached to, produced by, or evaluated during a policy decision. A capability can remain structurally non-executable until specified evidence conditions are satisfied. The resulting execution and proof then become part of the evidence state affecting subsequent executable authority.
Does AGT currently model this as a causal loop across ACS, AgentMesh, SRE, and its audit/evidence layers, or would evidence-conditioned execution authority be considered a separate governance primitive?
I'm experimenting with an implementation of this model and would be interested in how the AGT maintainers think about the boundary between evidence, authority, capability, and execution.
Derrick S.
All reactions