Save REA analysis evidence and read a small summary
Keep a full local analysis record, project a smaller summary, and write conclusions that separate observations, inferences, and unknowns. Commands checked against REA 6.4.0 documentation.
01 / Keep the record, reduce what you read
A large Electron app can produce more JSON than you can comfortably read or send to an AI client. Save the full result locally first, then use inspect-analysis-view to request a summary of that saved record. Separate CLI processes cannot reuse a connection-scoped MCP evidence reference.
These are documentation-based instructions, not an analysis run by this website. The paths are placeholders. The shell examples use a POSIX-compatible shell on macOS or Linux, or a compatible WSL environment, plus jq. This site accepts no application files.
Only analyze artifacts you are authorized to inspect. Static JavaScript analysis reads the target files; runtime capture is a separate operation that may execute the target. Local analysis does not determine what your AI client sends to its model provider.
02 / Save a version-specific analysis result
Use a fresh working directory with enough free disk space. Replace the target path and choose a new output filename: shell redirection can overwrite an existing file. Keep stderr and the exit status so a failed run cannot quietly become a successful report.
The command below pins rea-agents to 6.4.0 for this guide. Installation and compatibility must still be checked on your machine. Record the target version and a hash of the actual artifact or a file manifest alongside the JSON.
npx -y rea-agents@6.4.0 analyze-javascript-application "/absolute/path/to/app" --json > app-evidence.json 2> app-analysis.stderr.log
analysis_status=$?
printf "REA exit status: %s\n" "$analysis_status"Proceed only after reviewing the exit status and diagnostics. Status 0 means the operation completed; its result can still contain partial evidence, warnings, or unresolved questions. This page does not claim complete original-source recovery.
03 / Inspect a summary from the saved record
After a successful run, jq wraps the saved JSON in the inline evidence input expected by inspect-analysis-view. The next command runs only if jq successfully creates that input. Keep the full record; the smaller view is a reading aid.
To inspect module pages later, the official view contract also supports a page view with collection modules, an offset, and a limit. Consult the pinned workflow documentation for the full input contract instead of guessing fields.
jq -c '{source: {kind: "inline", evidence: .}, view: {kind: "summary"}}' app-evidence.json > app-view.json &&
npx -y rea-agents@6.4.0 inspect-analysis-view app-view.json --json > app-summary.jsonThis summary command is checked against official documentation, but has not been executed by this site against your app. Missing jq, invalid JSON, unsupported input, permissions, and resource limits remain possible failures.
04 / Write an evidence-bounded report
Start with one question, such as which module sends a clipboard IPC request. Attach each observation to a file, symbol, range, or evidence identifier that actually exists in your saved output. Explain the extra reasoning behind an inference, and keep unresolved behavior separate.
The copyable worksheet below is an original reporting aid. It is Markdown for your own notes, not a REA import schema and not an executed result.
# Analysis note — fill from your own record
Question: [one specific feature or behavior]
Target identity: [version + artifact hash or manifest]
REA version / host: [actual version and OS]
Run status / diagnostics: [exit code + warnings]
Saved record: [local filename; do not publish private paths]
Observation: [actual evidence + file/symbol/range]
Inference: [reasoning + confidence + alternatives]
Unknown: [missing evidence or unresolved behavior]
Next check: [specific static or authorized runtime check]
Outcome: [what that check actually showed]05 / Example: a clipboard clue is not a runtime result
Fictional teaching example: suppose your saved record contains a renderer module calling a clipboard IPC channel. That is an observation about the inspected files. It can support an inference that the feature crosses an IPC boundary. It does not by itself prove the handler exists in the installed build or that the clipboard write succeeds.
An unresolved handler or missing runtime capture stays unknown. A next check could locate the corresponding handler and its source range, then, if authorized, observe a declared copy action on a controlled target. Write the observed outcome separately. No clipboard app was analyzed or executed to produce this example.
- Do not turn a plausible call path into a claim of tested behavior.
- Do not interpret an absent observation as proof that a feature is absent.
- Before sharing a report, remove credentials, private code, and unnecessary local paths; keep the unredacted original securely on your machine.
Sources & version
This guide independently summarizes official documentation. It has not been verified against your environment or app. The official main branch may be ahead of the npm release.
REA 6.4.0 CLI, saved evidence, views and exit statusREA 6.4.0 JavaScript application workflow contractsREA 6.4.0 release