Tracework

Report first

Try a daily or weekly report right after install, even in conversation only. Use scoped facts already visible in the conversation before capturing; git-only coverage stays limited.

Close the day and week

Explain what changed, why it matters, and the next gate. Markdown brief by default; a presentation outline only when you ask.

Wrap up to keep the why

End key sessions by saving trade-offs, risks, and next steps so later reports stay grounded.

Trace progress to evidence

Follow report claims to raw facts, trade-offs, and verification evidence. Missing evidence stays visible.

How to use it

text

install -> try: write weekly / write daily
        -> configure vault and project groups when you want multi-day memory
        -> wrap up after key sessions
        -> generate daily / weekly / monthly reports when needed

Projects can declare a reporting group such as work or personal. Reports filter scope before selecting headlines, so personal projects never displace or leak into workplace output. A private all view keeps each group in a separate lane.

Boundary

Tracework is not a meeting-notes tool, approval workflow, performance packaging layer, generic office suite, or employee-monitoring surface. Activity volume is coverage, not proof of outcomes. When the record is thin, Tracework exposes the gap instead of inventing history.

Tracework is a local reporting plugin for Codex and Claude Code with five skills: capture keeps key facts; daily, weekly, and monthly produce progress reports; cold-start-interview configures storage and project groups.

Reports select projects before reading effective records, keeping work and personal groups separate. Monthly reviews use raw records and current risk states; Daily and Weekly reports provide prior judgments, so a complete Daily archive is optional. If some projects cannot be read, Tracework returns a conversation draft with coverage gaps and saves it only on explicit request.

TraceworkTurn agent work into evidence-backed progress reports.