Module evidence 

AI Evidence Definition

An Evidence captures the output of a single validation or quality assurance step — running tests, linting code, compiling the project, etc. It is the objective data that supports (or contradicts) the agent’s proposed changes.

Position in Lifecycle

⑥ ToolInvocation / ⑦ PatchSet
     │              │
     │              ▼
     └──────────▶ Evidence (run_id + optional patchset_id)
                        │
                        ▼
                    ⑨ Decision (verdict justification)

Evidence is produced during a Run, typically after a PatchSet is generated. The orchestrator runs validation tools against the PatchSet and creates one Evidence per tool invocation. A single PatchSet may have multiple Evidence objects (e.g. test + lint + build). Evidence that is not tied to a specific PatchSet (e.g. a pre-run environment check) sets patchset_id to None.

Purpose

• Validation: Proves that a PatchSet works as expected (tests pass, code compiles, lint clean).
• Feedback: Provides error messages, logs, and exit codes to the agent so it can fix issues and produce a better PatchSet.
• Decision Support: The Decision references Evidence to justify committing or rejecting changes. Reviewers can inspect Evidence to understand why a verdict was made.

How Libra should use this object

• Create one Evidence object per validation tool execution or report.
• Attach patchset_id when the validation targets a specific candidate diff.
• Use summary, exit_code, and report_artifacts for the durable audit record.
• Derive pass/fail dashboards and gating status in Libra; do not rewrite PatchSet or Run snapshots with validation summaries.

Structs

Evidence

Output of a single validation step (test, lint, build, etc.).

Enums

EvidenceKind

Kind of evidence.