Crash reports: replayable, bounded, local, reachable by agents (#384)

Every panic (in a command, in the kernel, anywhere else) and every unclean
exit (abort, stack overflow, kill) leaves a report in a local crash folder:
message, file:line, a trimmed backtrace, version, commit, system and renderer,
the command and recent commands, and when possible the design and replay.json
that reproduce it (solvecraft-cli crashes replay <id>).

- Cost: nothing until the app installs the hook (tests, CLI and web never
  do); then one small journal line per top-level command of the design on
  screen, taken back with a truncate when the command changed nothing.
- Unclean exits: each run holds a lock on its journal; a journal nobody holds
  at the next start belongs to an app that died. Autosaves restart the
  journal from the autosaved design.
- Bounded: a crash's signature (place + message, numbers ignored) is counted,
  not rewritten; at most 5 new reports per run, 20 reports, 50 MB, 90 days;
  recovered kernel panics are pruned first.
- Agents: diagnostics.crash_reports / crash_report / crash_reports.clear
  commands, the MCP crash_reports tool, solvecraft-cli crashes, solvecraft
  --crash-log; diagnostics.crash_test (with SOLVECRAFT_CRASH_TEST) checks it.
- People: the Crash Reports window opens after a crash (Help > Crash
  Reports…) with a reviewed issue draft per report; nothing is sent.
- Release builds keep line tables for file:line in backtraces.

Tests: child processes that really panic in a command, in the kernel, on a
thread, and abort; dedup counts, limits, hostile folders, journal parsing,
the UI window and draft, the MCP tool. docs/crash-reports.md.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>

crates/engine/src/cmd/diagnostics.rs

@@ -0,0 +1,74 @@
+//! Crash reports, for agents and people (see `crate::crash` and docs/crash-reports.md): list
+//! them, read one, delete them. They live on this computer and are never sent anywhere.
+
+use serde_json::{Value, json};
+
+use super::CommandSpec;
+use crate::params::{bad, str_};
+use crate::{EngineError, Result, Session, crash};
+
+pub static COMMANDS: &[CommandSpec] = &[
+    CommandSpec::new("diagnostics.crash_reports", "Crash Reports", list).noundo().params(
+        "limit?: number (default 20). Lists this computer's crash reports, newest first: {dir, reports: [{id, kind, time, message, \
+         command, count, version, path, replayable}]}. kind: command_panic | panic | kernel_panic_recovered | unclean_exit. \
+         Read one with diagnostics.crash_report; reproduce it with `solidcraft-cli crashes replay <id>`",
+    ),
+    CommandSpec::new("diagnostics.crash_report", "Crash Report", read).noundo().params(
+        "id: report id. The whole report: message, location, backtrace, app version and commit, system, the command that \
+         crashed and the recent ones, notes, and design_path / replay_path when the crash can be replayed \
+         (`solidcraft-cli run <replay_path> --in <design_path>`)",
+    ),
+    CommandSpec::new("diagnostics.crash_reports.clear", "Delete Crash Reports", clear)
+        .noundo()
+        .params("id?: one report (default: all). Returns {deleted}"),
+    CommandSpec::new("diagnostics.crash_test", "Crash Test", crash_test).noundo().enabled(crash_test_enabled).params(
+        "mode: panic | kernel | abort. Checks that crash reports work: panic = an internal error in this command (caught; \
+         the design is unchanged), kernel = one inside the geometry kernel (caught), abort = the app ends at once (reported \
+         at the next start). Only when the SOLVECRAFT_CRASH_TEST environment variable is set",
+    ),
+];
+
+fn dir(cmd: &str) -> Result<std::path::PathBuf> {
+    crash::default_dir().ok_or_else(|| bad(cmd, "no crash folder on this system"))
+}
+
+fn list(_: &mut Session, p: &Value) -> Result<Value> {
+    let d = dir("diagnostics.crash_reports")?;
+    let limit = p.get("limit").and_then(Value::as_u64).map_or(20, |n| n.min(1000)) as usize;
+    let mut reports = crash::list(&d);
+    reports.truncate(limit);
+    Ok(json!({ "dir": d.to_string_lossy(), "reports": reports }))
+}
+
+fn read(_: &mut Session, p: &Value) -> Result<Value> {
+    let cmd = "diagnostics.crash_report";
+    let id = str_(p, "id").ok_or_else(|| bad(cmd, "`id` is required (see diagnostics.crash_reports)"))?;
+    crash::read(&dir(cmd)?, id).map_err(|e| bad(cmd, e))
+}
+
+fn clear(_: &mut Session, p: &Value) -> Result<Value> {
+    let d = dir("diagnostics.crash_reports.clear")?;
+    Ok(json!({ "deleted": crash::clear(&d, str_(p, "id")) }))
+}
+
+fn crash_test_enabled(_: &Session) -> std::result::Result<(), String> {
+    if std::env::var_os("SOLVECRAFT_CRASH_TEST").is_some() { Ok(()) } else { Err("set SOLVECRAFT_CRASH_TEST to use the crash test".into()) }
+}
+
+// The one deliberate panic: it proves the crash reports work, and runs only when asked for with
+// SOLVECRAFT_CRASH_TEST.
+#[allow(clippy::panic)]
+fn crash_test(_: &mut Session, p: &Value) -> Result<Value> {
+    let cmd = "diagnostics.crash_test";
+    match str_(p, "mode") {
+        Some("panic") => std::panic::panic_any(String::from("crash test: an internal error in a command")),
+        Some("kernel") => {
+            solvecraft_kernel::guard("crash test", || -> solvecraft_kernel::Result<()> {
+                std::panic::panic_any(String::from("an internal error in the kernel"))
+            })?;
+            Err(EngineError::Other("crash test: the kernel error was not caught".into()))
+        }
+        Some("abort") => std::process::abort(),
+        _ => Err(bad(cmd, "`mode` must be panic, kernel or abort")),
+    }
+}

crates/engine/src/cmd/mod.rs

@@ -16,6 +16,7 @@ mod component_clone;
 mod component_clone_assembly;
 mod construct;
 mod delete;
+mod diagnostics;
 mod edit;
 mod face;
 mod features;
@@ -214,6 +215,7 @@ pub fn command_specs() -> Vec<&'static CommandSpec> {
     v.extend(file::COMMANDS.iter());
     v.extend(inspect::COMMANDS.iter());
     v.extend(section::COMMANDS.iter());
+    v.extend(diagnostics::COMMANDS.iter());
     v
 }