Improve agent-facing history endpoints (Phase 1)

- Fix /api/history/{type} to search the full persistent store instead of
  only the bounded in-memory ring buffer, so filtered history isn't
  silently truncated.
- Add structured type/since/until filters to /api/history/search and
  /api/history/wait, alongside the existing free-text query.
- Add GET /api/traces/{traceId} to fetch all entries for a trace in one
  call instead of relying on substring search.
- Rate-limit /api/history/wait per client IP.
- Add docs/PLAN.md tracking the phased improvement plan; update README
  and agents_example.md for the new/changed endpoints.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

docs/PLAN.md

@@ -0,0 +1,44 @@
+# Improvement plan
+
+Priority order: (1) agent endpoints, (2) recent issues for the web UI, (3) a
+Sentry-like frontend error collector.
+
+## Phase 1 — Agent endpoints (highest priority)
+
+- [x] Fix `HandleHistoryByType`: it only filtered the in-memory ring buffer
+      (capped at `historySize`, default 100) instead of the persistent store,
+      so `/api/history/traces|metrics|logs` silently returned far less than
+      `/api/history/search` could see. Now routed through `SearchHistory`.
+- [x] Structured filters on `/api/history/search` and `/api/history/wait`:
+      `type=`, `since=`, `until=` (unix nanoseconds), combinable with the
+      existing free-text `q=`.
+- [x] `GET /api/traces/{traceId}` — assemble all entries referencing one
+      trace id in a single call, instead of relying on substring search
+      matching the id inside raw JSON.
+- [x] Basic per-IP rate limiting on `/api/history/wait` (a blocking
+      long-poll, cheap to abuse).
+- [ ] Cursor-based pagination for `/api/history/search` (deferred).
+- [ ] Machine-readable tool manifest (`/.well-known/agent-tools.json` or an
+      MCP server) replacing `agents_example.md` (deferred).
+
+## Phase 2 — Recent issues for the web UI
+
+- [ ] Fingerprint spans with `status=ERROR`/exception events and logs with
+      `severity >= ERROR` into deduped Issues (count, firstSeen, lastSeen,
+      sample entries, affected services), updated incrementally on ingest.
+- [ ] `GET /api/issues`, `GET /api/issues/{id}`.
+- [ ] UI "Issues" tab (lives in the separate frontend/`dist` project).
+
+## Phase 3 — Frontend error collector (Sentry-style)
+
+- [ ] `static/otel-rum.js`: captures `window.onerror`,
+      `unhandledrejection`, resource load failures; translates them to OTLP
+      log records client-side and POSTs to the existing `/v1/logs` endpoint
+      (so Phase 2's issue derivation picks them up for free via a
+      `source: frontend` attribute).
+- [ ] CORS support on the OTLP HTTP endpoints (currently absent — required
+      for any browser-origin POST).
+- [ ] Lightweight ingest key for scoping/rate-limiting noisy pages (not
+      real auth).
+- [ ] Explicitly out of scope for v1: source map resolution, session
+      replay, PII scrubbing.