v0.2.2

What's new

Richer replay output

The replay now shows what the agent actually did, not just event type labels.

Before:

+  6.06s → tool_call Bash (command, description)
+ 27.65s ✗ error
+ 46.01s ← tool_result  (51ms)
+1m52.54s 🤖 assistant_response
              "**75 tests**, all passing in `3.60s`."

After:

+  6.06s → tool_call Bash
              $ python -m pytest tests/ -v 2>&1
+ 27.65s ✗ error Bash
              Command failed with exit code 1
+ 46.01s ← tool_result Bash (51ms)
              75 tests, all passing in 3.60s
+1m52.54s 🤖 assistant_response
              "75 tests, all passing in 3.60s."

• tool_call shows actual values: Bash commands, file paths, glob patterns, grep queries, URLs, agent prompts
• tool_result shows tool name and output preview
• error shows tool name and error message (not just "error")
• assistant_response strips markdown before display

Install / upgrade

uv tool upgrade agent-strace
pip install agent-strace==0.2.2

101 tests. Zero dependencies. Python 3.10+ stdlib only.