Learner activity report

Export what you did in a portable activity, a snapshot of your experiment journal, or both, as a local report you can open in a browser, check and choose to share. The export checks that the records are well formed and in order. It does not grade you, infer what you learned or validate the science.

Two record types can go in:

  • Learner actions: a sidecar file written by a portable activity. It separates actions the browser observed from assistance the harness declared. A missing assistance event means coverage is unknown, not that no help occurred.
  • Journal snapshot: a copy of your experiment journal. It is not authenticated. When both are selected, the activity's scene id must appear in the journal; that overlap still does not authenticate a learner, lesson or session, and the two clocks stay separate.

The activity-to-report tutorial walks through a full example.

Check the actions file

From the repository root:

.venv/bin/python tools/learner_trace.py validate path/to/learner-actions.json
.venv/bin/python tools/learner_trace.py schema --out learner-actions.schema.json

The JSON Schema is generated from the same model the exporter uses, so producers and the report share one contract.

Export

Export the actions with the exact prepared activity they came from:

.venv/bin/python tools/learner_report.py export reports/learner-report \
  --actions path/to/learner-actions.json \
  --activity path/to/prepared-activity/activity.json

The sidecar's activity_sha256 must match the digest of that prepared activity, so labels, source statement, parameters, choices and controls in the report belong to the captured definition. Without --activity, the report marks the context as unavailable instead of guessing labels.

To include Add
A journal snapshot alongside the actions --journal path/to/journal.json
A journal alone --journal without --actions
Your original free text --include-original-text

Free text and journal evidence locators are redacted by default. Keeping your own words is an explicit local choice. Consent to store the journal is not consent to share or publish the report. The exporter does not copy learner-identity or private-session fields, but activity ids and source descriptions can still identify you: read the report before sharing it. Redaction is a policy over the allowed fields, not a general de-identification guarantee.

Open and verify

Open reports/learner-report/index.html: a portable, escaped view showing the activity and the journal as separate timelines, with links to the ASTRA, ARA, MyST and integrity files (see the glossary for these formats).

To view it in the MySTRA Observatory, point the course root at the report's parent folder:

.venv/bin/python services/observatory/observatory_app.py --course-root reports --port 8124

Then open http://127.0.0.1:8124/courses/learner-report/mystra. The page shows the redacted actions and journal events. Its summary is course-oriented and shows zero scenes judged: expected, since assessment is outside the export. More on the Observatory: observatory.

Check an export before relying on it:

.venv/bin/python tools/learner_report.py verify reports/learner-report

Verification revalidates the selected sources and their bindings, recomputes the summary and the rendered outputs, runs the installed astra-tools validator (ASTRA 0.0.14) and checks the SHA-256 hash of every standard output and selected source. The manifest is unsigned: it detects stale, missing or accidentally changed files, but someone who can rewrite the whole report can rewrite its hashes too.

What the folder contains

File Content
sources/learner-actions.json, sources/experiment-journal.json, sources/activity.json Only the selected, redacted records
report.json The derived technical summary and its scope
astra.yaml The ASTRA 0.0.14 artifact declaration
trace/exploration_tree.yaml Separate, evidence-linked action journeys
course_provenance.myst.md Output of the shared MySTRA renderer
index.html The portable view
checks.json Validator result and file hashes (it does not hash itself)

The folder never contains a copy of the original raw trace. The code is in tools/learner_report.py and tools/learner_trace.py.

View the source of this page on GitHub