See what was achieved and what remains.

Turn scattered execution records into a retrospective.

Review deliverables, status, and outstanding work for a project and period.

To understand rework, waiting, or human intervention, examine process evidence and identify supported causes and improvements.

When an old report has omissions or classification errors, add evidence, correct the ledger, and preserve the original report.

Review outcomes first. Examine the process when needed.

Choose a day, week, month, quarter, year, or explicit date range. Request standard or deep analysis separately; a longer period does not automatically mean deeper analysis.

Define the projects and the period first.

Specify the project scope and reporting period. For a cross-project report, list the projects and choose a shared report directory.

Relatra needs Python 3.9+ and readable local session logs or supported workflow evidence. Use it independently or to review existing Cohestra and Cadra execution records.

For comparisons with earlier periods, provide archived reports with matching scope and comparable measurement definitions.

Review the last complete week first.

After installation, enter this in an agent chat that supports skills:

$relatra weekly

Project: <absolute path to your project>
Report directory: <absolute path to the report directory>
Report language: English
Time zone: Asia/Shanghai

Without explicit overrides, Relatra defaults to Chinese and Asia/Shanghai, covering the previous complete calendar week from Monday through Sunday. This example explicitly requests English while retaining that time zone.

A standard report summarizes outcomes, status, outstanding work, and observed time and tokens. Ask explicitly to analyze rework, human intervention, or waiting.

For current progress, say “this week so far.” The report will identify its actual cutoff time.

This is an agent-chat example. Replace <…> with actual values.

Keep conclusions concise and evidence traceable.

Reports highlight scope, verified outcomes, outstanding work, and key resource observations. Detailed records remain in report data; necessary evidence limitations also appear in the prose.

Later retrospectives can reuse the ledger and focus on new activity, changed evidence, and unresolved goals.

Historical corrections produce a new version and preserve the earlier report.

State what can be confirmed. Leave missing evidence explicit.

Verify outcomes and status against actual records. Engineering deliverables without readable evidence retain a verification requirement.

Time and tokens come from readable records; missing values remain unknown. Time analysis explains its measurement basis and does not convert usage into costs, labor hours, or time saved.

Coverage depends on available records. Cloud activity, deleted records, and work on other machines may be absent. The first collection also states historical coverage limits.

Relatra is for retrospectives. It does not monitor work live or automatically execute the original tasks. Their status changes and follow-up actions belong to the corresponding workflow.

Choose a period and inspect the outcomes already recorded.

Start with one project and a defined scope. Confirm what was achieved and what remains, then examine the process if needed.

Need to continue the engineering work identified in the report?

Explore Cohestra →