Post-Analysis

Run analyzers on completed result directories.

Use post-analysis when a run has completed or partially completed and you need analyzer output without rerunning the agent.

Command

agentcompass analysis \
  --input results/swebench_verified/your-model/20260703_120000 \
  --analysis-params '{
    "analyzers": ["ExceptionAnalyzer", "TruncationAnalyzer"]
  }'

By default, analysis copies the input run into a new timestamped sibling. Use --override only when you intentionally want to mutate the original run directory.

Qualitative Analysis

agentcompass analysis \
  --input results/terminal_bench_2/your-model/20260703_120000 \
  --analysis-params '{
    "analyzers": ["QualitativeAnalyzer"],
    "QualitativeAnalyzer": {"render_mode": "file"}
  }' \
  --task_concurrency 8

Outputs

Artifact	Purpose
analysis_result.<AnalyzerId>	Per-task analyzer payload in detail files.
analysis_summary.json	Machine-readable aggregate analyzer output.
analysis_summary.md	Human-readable analyzer summary.

Related Pages

• Analyzers
• Analyzer reference
• Results