agentv 3.11.1 → 3.13.0

package/README.md

@@ -203,14 +203,17 @@ See `agentv eval --help` for all options: workers, timeouts, output formats, tra
 Write results to different formats using the `-o` flag (format auto-detected from extension):
 
 ```bash
-# JSONL (default streaming format)
-agentv eval evals/my-eval.yaml -o results.jsonl
+# Default run workspace (index.jsonl + benchmark/timing/per-test artifacts)
+agentv eval evals/my-eval.yaml
 
 # Self-contained HTML dashboard (opens in any browser, no server needed)
 agentv eval evals/my-eval.yaml -o report.html
 
+# Explicit JSONL output
+agentv eval evals/my-eval.yaml -o output.jsonl
+
 # Multiple formats simultaneously
-agentv eval evals/my-eval.yaml -o results.jsonl -o report.html
+agentv eval evals/my-eval.yaml -o report.html
 
 # JUnit XML for CI/CD integration
 agentv eval evals/my-eval.yaml -o results.xml
@@ -218,10 +221,13 @@ agentv eval evals/my-eval.yaml -o results.xml
 
 The HTML report auto-refreshes every 2 seconds during a live run, then locks once the run completes.
 
-You can also convert an existing JSONL results file to HTML after the fact:
+By default, `agentv eval` creates a run workspace under `.agentv/results/raw/<run>/`
+with `index.jsonl` as the machine-facing manifest.
+
+You can also convert an existing manifest to HTML after the fact:
 
 ```bash
-agentv convert results.jsonl -o report.html
+agentv convert .agentv/results/raw/eval_<timestamp>/index.jsonl -o report.html
 ```
 
 #### Timeouts
@@ -352,7 +358,7 @@ agentv create eval my-eval          # → evals/my-eval.eval.yaml + .cases.jsonl
 Compare a combined results file across all targets (N-way matrix):
 
 ```bash
-agentv compare results.jsonl
+agentv compare .agentv/results/raw/eval_<timestamp>/index.jsonl
 ```
 
 ```
@@ -373,8 +379,8 @@ Pairwise Summary:
 Designate a baseline for CI regression gating, or compare two specific targets:
 
 ```bash
-agentv compare results.jsonl --baseline gpt-4.1                          # exit 1 on regression
-agentv compare results.jsonl --baseline gpt-4.1 --candidate gpt-5-mini  # pairwise
+agentv compare .agentv/results/raw/eval_<timestamp>/index.jsonl --baseline gpt-4.1
+agentv compare .agentv/results/raw/eval_<timestamp>/index.jsonl --baseline gpt-4.1 --candidate gpt-5-mini
 agentv compare before.jsonl after.jsonl                                  # two-file pairwise
 ```
 
@@ -518,10 +524,7 @@ tests:
 
 Scoring: `(satisfied weights) / (total weights)` → verdicts: `pass` (≥0.8), `borderline` (≥0.6), `fail`
 
-Auto-generate rubrics from expected outcomes:
-```bash
-agentv generate rubrics evals/my-eval.yaml
-```
+Author assertions directly in your eval file. When you want help choosing between simple assertions, deterministic graders, and LLM-based graders, use the `agentv-eval-writer` skill.
 
 See [rubric evaluator](https://agentv.dev/evaluation/rubrics/) for detailed patterns.
 

package/dist/{agentv-provider-MIDKLYIH-6LIYKQRP.js → agentv-provider-NFFLXG5M-TJAWCWCX.js}

@@ -2,9 +2,8 @@ import { createRequire } from 'node:module'; const require = createRequire(impor
 import {
   AgentvProvider
 } from "./chunk-BJV6MDBE.js";
-import "./chunk-UGXG73VF.js";
 import "./chunk-5H446C7X.js";
 export {
   AgentvProvider
 };
-//# sourceMappingURL=agentv-provider-MIDKLYIH-6LIYKQRP.js.map
+//# sourceMappingURL=agentv-provider-NFFLXG5M-TJAWCWCX.js.map