Daily Ops 1.1.0: review decisions and preview changes

Daily Ops 1.1 makes everyday plans easier to review and revise. It keeps the existing local task format and planning rule while adding a complete proposal-to-preview workflow.

• HTML reports lead with the first task, capacity arithmetic, and due or waiting work that needs a decision.
• Each visible task starts unreviewed. Explicitly keep it or propose an estimate, deadline, deferral, or waiting-status change. Download changes.json and review.md to retain your choices.
• preview FILE --output reports/change.html --date YYYY-MM-DD --minutes N shows exact task changes and the actual before/after plans without saving the proposal.
• lint-plan FILE compares canonical plan JSON with current workspace state, catching altered facts and stale revisions while retaining useful advisories.
• The shared skill explains the same workflow for Claude, Codex, and capable ChatGPT environments, with a manual fallback when runtime access is unavailable.

Try the generated plan · See its change preview · Install · Executed evidence

All five release-commit CI jobs passed across Linux, macOS, and Windows. Validation includes 102 local tests, executable worksheet JavaScript checks, real browser downloads passed into preview/apply, stale-revision refusal, independent code review, a separate assistant skill trial, and verified release archives. The HTML was checked at a 320-pixel viewport, with enlarged text, without JavaScript, and under print emulation. These checks demonstrate software behavior, not measured human productivity gains or complete accessibility conformance.

Draft report choices are held only in page memory and are lost on reload unless downloaded. The report cannot apply tasks; that still requires Python 3.10+ or an assistant with runtime and workspace access. Native GUI/model execution on every host is not established. No new runtime dependencies, task-state migration, or telemetry are introduced.

Choose the standalone skill ZIP, Claude plugin ZIP, or full source ZIP below. SHA256SUMS covers all three archives. The source includes the fictional fixture and generator for rebuilding the public examples. Public html-plan inspiration is credited in the design notes; the implementation is independently written.

Daily Ops 1.0.0

Daily Ops helps turn a task list into a realistic plan for the time available. Work that does not fit, waits on someone else, or needs a deadline decision remains visible.

This first release includes:

• A shared assistant skill for Claude and Codex, with documented ChatGPT installation and manual exchange paths.
• A local Python runtime with no runtime dependencies, persistent task IDs, previewable changes, and stale-revision protection.
• Explainable time-budget planning, completion review, portable exports, and self-contained HTML reports.
• An interactive browser demo with fictional examples for students, freelancers, caregivers, and team contributors.

Try the live demo · Installation instructions · Validation evidence and limitations

Choose the Claude plugin ZIP for a supported plugin upload workflow, the skill ZIP for a supported standalone skill installation, or the source ZIP for the complete public source and local command-line tool. Python 3.10 or later is required for the runtime. Verify downloads against SHA256SUMS.

Validation includes 62 local tests, a five-job Linux/macOS/Windows CI matrix, native Claude Code and Codex CLI package-installation checks, an independent explicitly invoked agent trial, browser interaction checks, and reproducible archive verification. Native GUI installation and automatic triggering have not been tested on every host. No human productivity or time-savings study has been performed.

MIT licensed. The runtime contains no telemetry, account connectors, or background uploads. Content provided to an AI host remains subject to that provider's processing and terms.

Release verification on October 7, 2026: all five CI jobs passed on the tagged commit, and the matching site deployment succeeded. An independent check downloaded all four release assets without authentication, verified every ZIP against SHA256SUMS, and ran all 62 tests successfully from the downloaded source. A fresh-workspace smoke test selected a 30-minute task within a 40-minute budget, visibly deferred a 20-minute task, generated the report, and confirmed that planning left saved task state unchanged.

A separate macOS smoke test denied all network access with an operating-system sandbox and supplied only PATH and LANG environment variables. Capture, planning, HTML report generation, completion, copying the workspace, and resuming succeeded; the sandbox independently rejected a network probe with EPERM.