Should you try PanWatch?
Evaluate PanWatch for combined portfolio monitoring and configurable AI research if you can verify exports and integrations on your intended deployment. Our September 21, 2026, sandbox run recorded two PDF text-extraction assertion failures whose cause remains unresolved. Those results justify checking reports; they do not establish a defect in current PanWatch.
The current README documents an English interface alongside Simplified Chinese. It describes holdings, watchlists, compound price alerts, scheduled research, paper trading and a mobile PWA. English-only readers should not be excluded because the historical README was Chinese.
What the stored run established
The recorded run tested commit 89bdf3f, associated with release 0.14.0, on September 21, 2026. The sandbox had 3 CPUs, 8 GB of memory and a Python 3.12 Bookworm image. This correction uses that stored run; no new execution was performed.
Install and build recorded exit 0; tests recorded exit 1. The harness durations were 31 seconds for install, 1 second for build and 34 seconds for tests. The dependency footprint was 172 packages and 595 MB.
The install tail reports editable-build and wheel-build failures, including an invalid distribution name or version. The supplied evidence does not explain why the overall install step returned zero. The build tail is empty, so its successful status does not establish a running web application or a completed production workload.
The stored parser summaries record 782 passed, 2 failed, total 784 and 3 skipped. These fields do not establish exhaustive unique test coverage. The raw pytest footer reports 2 failed, 782 passed and 3 skipped in 29.26 seconds. That suite duration differs from the harness's 34-second test-step duration; neither replaces the other. Skipped tests are not failures.
MrKeyoor's testing methodology explains the recording approach and its limits. Its supplied current description does not establish which methodology revision governed this September run.
What the PDF assertions mean
Both failed assertions are in tests/test_pdf_export.py. The markdown-to-PDF test could not find the expected Chinese company name, 广汽集团, in extracted PDF text. The endpoint test could not find its expected full-analysis marker, 技术面分析正文UNIQUE, in text extracted from the PDF attachment.
The stored run recorded two PDF text-extraction assertion failures whose cause remains unresolved. Missing expected strings do not establish whether the cause was font handling, extraction, report construction or the sandbox environment. They also do not establish whether current PanWatch behaves the same way.
Before depending on exports, generate a report on your intended deployment, search and copy its Chinese text, and compare the attachment with the complete analysis shown in the application. These are recommended acceptance checks, not work completed by the sandbox. The recorded run did not verify live quotes, model responses, scheduled delivery, brokerage synchronization or trading returns.
The stored audit reports zero advisories and packages: 0. The supplied evidence does not explain that package count or establish meaningful dependency scan coverage; this result cannot support a security conclusion.
What remains to configure
The documented Docker setup exposes port 8000 and persists /app/data. Set the deployment timezone explicitly for background schedules and notification quiet hours. Chromium's first-start download requires network access unless browser installation is skipped. Model requests and market-data collection also depend on the services you configure.
Read the README pinned to the measured commit when evaluating that version. It lists Python 3.10+, Node.js 18+ and pnpm for development. Current development instructions describe different requirements and should not be treated as instructions tested in the historical run.
Start with one holding and one analysis. Check the model connection, compare the report with its sources and test one notification channel before enabling scheduled workflows. A successful command status alone does not settle these deployment checks.
Alternatives for different workflows
TradingAgents is the research framework PanWatch integrates. TradingAgents documents CLI and Python package use. Consider it directly when developing or studying the research pipeline is your main job. The supplied evidence does not establish it as a replacement for every PanWatch dashboard feature.
Ghostfolio documents transaction management, portfolio performance and transaction import and export. Consider it when maintaining investment records is your main job. It can complement a separate research tool. Neither alternative was installed or compared in the recorded PanWatch run.
How to read the ratings
The numerical ratings are editorial judgments about the supplied evidence, not benchmarks. For setup, documentation and maturity, 1 means substantial documented barriers, 2 means major limitations, 3 means material qualifications, 4 means strong corroborated evidence within that dimension and 5 means broad corroboration across deployments or versions. Missing evidence remains unassessed.
Setup ease is 3: the recorded command statuses coexist with installation errors, and complete deployment remains unverified. Documentation is 3: the supplied README covers setup and configuration, but current instructions must be kept separate from the measured version. Maturity is 3: the stored run recorded two PDF text-extraction assertion failures whose cause remains unresolved, and newer documented behavior has no matching sandbox run in the supplied evidence.
Community is unassessed. Repository attention figures do not establish adoption, contributor support or response quality, and the supplied evidence does not support a numerical judgment of those qualities.