docs: add missing event data fields so API users see what the engine actually emits

CHECK_PASSED/CHECK_FAILED events include `output`, iteration events include
`result_text`, and CHECKS_COMPLETED results include `output` — all present in
the code but missing from the docs. Also removes a stale "planned dashboard"
reference from the RunConfig docstring.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: 2 people

docs/api.md

@@ -224,18 +224,18 @@ All event types the loop can emit. Every event has `type`, `run_id`, `timestamp`
 | Event type | When | Data fields |
 |---|---|---|
 | `ITERATION_STARTED` | Iteration begins | `iteration` |
-| `ITERATION_COMPLETED` | Agent exits with code 0 | `iteration`, `returncode`, `duration` (seconds), `duration_formatted`, `detail`, `log_file` |
-| `ITERATION_FAILED` | Agent exits non-zero | `iteration`, `returncode`, `duration`, `duration_formatted`, `detail`, `log_file` |
-| `ITERATION_TIMED_OUT` | Agent exceeds timeout | `iteration`, `returncode` (null), `duration`, `duration_formatted`, `detail`, `log_file` |
+| `ITERATION_COMPLETED` | Agent exits with code 0 | `iteration`, `returncode`, `duration` (seconds), `duration_formatted`, `detail`, `log_file`, `result_text` |
+| `ITERATION_FAILED` | Agent exits non-zero | `iteration`, `returncode`, `duration`, `duration_formatted`, `detail`, `log_file`, `result_text` |
+| `ITERATION_TIMED_OUT` | Agent exceeds timeout | `iteration`, `returncode` (null), `duration`, `duration_formatted`, `detail`, `log_file`, `result_text` |
 
 #### Checks
 
 | Event type | When | Data fields |
 |---|---|---|
 | `CHECKS_STARTED` | Check phase begins | `iteration`, `count` |
-| `CHECK_PASSED` | A single check passes | `iteration`, `name`, `passed`, `exit_code`, `timed_out` |
-| `CHECK_FAILED` | A single check fails | `iteration`, `name`, `passed`, `exit_code`, `timed_out` |
-| `CHECKS_COMPLETED` | All checks finish | `iteration`, `passed`, `failed`, `results` (array of `{name, passed, exit_code, timed_out}`) |
+| `CHECK_PASSED` | A single check passes | `iteration`, `name`, `passed`, `exit_code`, `timed_out`, `output` |
+| `CHECK_FAILED` | A single check fails | `iteration`, `name`, `passed`, `exit_code`, `timed_out`, `output` |
+| `CHECKS_COMPLETED` | All checks finish | `iteration`, `passed`, `failed`, `results` (array of `{name, passed, exit_code, timed_out, output}`) |
 
 #### Prompt assembly
 

src/ralphify/_run_types.py

@@ -40,7 +40,6 @@ class RunConfig:
     Mutable by design: the engine reads fields at each iteration boundary,
     so you can change ``max_iterations``, ``delay``, or ``timeout`` while
     the loop is running and the new values take effect on the next cycle.
-    This is how the planned dashboard will let users tune a live run.
 
     For CLI usage the config is built once from ``ralph.toml`` + flags.
     For programmatic usage, construct directly and pass to :func:`run_loop`.