Daily work planner
Skill Stephen-studying/daily-work-planner/daily-work-planner
Local-first work-session preflight skill for Codex, Claude Code, Gemini CLI, and Agent Skills clients.
npx -y skills add Stephen-studying/daily-work-planner --skill daily-work-plannerAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Plan a focused work session before execution begins, inspect current/local open tasks, estimate required time, score feasibility against the actual deadline window, checkpoint progress, resume unfinished sessions, create handoffs, and record private local task memory. Use when the user asks to start, schedule, estimate, resume, or replan a work session with goals, files, available time, deadlines, current-window notes, or a local repository; support milestones, acceptance criteria, buffers, fallback scope, repo task triage, speed-adjusted estimates, feasibility scoring, TXT/DOCX/ICS outputs, review logs, and habit memory for documents, code, study, research, presentations, data, or project work. Do not invoke merely because files or a repository are present when the user wants the substantive task executed rather than planned.
SKILL.md
8.0 KB, as published. Nobody here has run it
Daily Work Planner
This is the Codex/OpenAI skill entrypoint. For agents that read generic instruction files instead of SKILL.md, use AGENTS.md in this same folder.
Core Rule
Optimize for execution, not planning. Convert uncertainty into a usable first plan quickly, then use checkpoints to correct the plan during work.
Before planning, calculate the planning budget:
planning_budget = min(available_work_time * 5%, absolute_cap)
Use these caps:
- <= 60 minutes total work: max 3 minutes planning
- 61-240 minutes total work: max 8 minutes planning
- 241-480 minutes total work: max 15 minutes planning
- Multi-day or > 480 minutes: max 20 minutes for the first planning pass
If the planning budget is small, produce a minimal plan instead of asking many questions.
Workflow
- Identify the task goal, current-window task text if present, available time, hard deadline, expected output, local repo, and provided files.
- Estimate required minutes separately from available minutes. If a hard deadline exists, calculate the actual start-to-deadline window; never substitute the required-time estimate for available time.
- Score feasibility by comparing the actual available window with the required-time estimate. Use
scripts/feasibility_score.pyto choose complete, deliverable, minimum, or redefine scope. - Calculate planning budget and choose output depth: quick, standard, or full.
- Inspect only enough source material to determine scope and priority. Use
scripts/extract_file_context.pyfor local PDFs, DOC/DOCX, PPTX, XLSX, Markdown, text, and code when deterministic file context is useful. - Choose or classify the work mode. Use
scripts/classify_work_mode.pywhen the mode is unclear. - Classify files as primary, reference, optional, or ignore-for-now. Use
scripts/rank_files.pywhen multiple files are provided. - Set hard deadline, soft deadline, buffers, and checkpoint times.
- Split the work into milestones with acceptance criteria.
- Define the minimum deliverable version.
- Add a delay-response plan for checkpoint slippage.
- During execution, use
scripts/checkpoint_session.pyto record progress and detect whether the plan should be compressed. - If the user resumes later, use
scripts/resume_session.pyto create a concise resume card before replanning. - At the end, use
scripts/handoff_session.pyto create a handoff and optionally write local memory. - Validate the plan with
scripts/validate_plan.pywhen a durable plan is being saved or shared. - For complete work-session packages, persist
session.jsonand update it throughscripts/session_state.py. - After completion or review, record a sanitized local memory entry with
scripts/task_memory.pywhen the user wants task history, estimate calibration, or habit memory. Support list, forget, purge, retention, and export operations. - Output the plan in a compact form the user can execute immediately.
Source Handling
Respect the planning budget when reading files.
- For large PDFs or documents, inspect titles, headings, tables of contents, abstracts, filenames, metadata, and representative pages first.
- Avoid deep summarization unless the user's task is explicitly reading or analysis.
- Prefer local file inspection. Do not upload sensitive files.
- If files are too many, prioritize by filename, modified time, user-stated relevance, and relationship to the final deliverable.
- Do not claim to read private OS windows directly. In Codex, use the current conversation/context that is already available; in scripts, accept copied window text through
--window-note.
Output Depth
Use Quick Plan for short or unclear tasks:
- Goal
- Planning budget
- First action
- 2-4 steps
- Soft and hard deadline
- Minimum deliverable
Use Standard Plan for most same-day work:
- Task summary
- Planning budget
- File priority
- Soft and hard deadline
- Milestone table
- Buffer
- Minimum deliverable
- Delay response
Use Full Plan only for complex full-day or multi-day work:
- Scope decision
- File priority table
- Milestones and acceptance criteria
- Checkpoints
- Buffer and fallback paths
- Todo list
- Review log template
- Optional ICS event instructions
- Consolidated TXT/DOCX report for human reading
References
Load these only when needed:
references/planning_rules.md: budget, soft deadline, buffer, and output-depth rules.references/work_modes.md: work-mode-specific planning rules.references/reschedule_rules.md: checkpoint delay handling and minimum deliverable rules.references/privacy_rules.md: local-first and sensitive-content handling.references/memory_rules.md: local task memory, habit learning, and what not to store.
Scripts
Use scripts when deterministic output is helpful:
scripts/plan_day.py: create a Markdown work-session plan from start time, total minutes, mode, goal, and optional deadline.scripts/extract_file_context.py: extract lightweight local context from PDF, DOC/DOCX, PPTX, XLSX, Markdown, text, and code files for priority decisions.scripts/inspect_tasks.py: inspect current-window notes and local repositories for open tasks, then estimate speed-adjusted required minutes.scripts/feasibility_score.py: score whether the work fits the time box and choose complete, deliverable, minimum, or redefine scope.scripts/classify_work_mode.py: infer the most likely work mode from the goal and file names.scripts/rank_files.py: rank files as primary, reference, optional, or ignore-for-now.scripts/validate_plan.py: check that a plan includes deadlines, milestones, acceptance criteria, buffer, fallback, and budget compliance.scripts/start_session.py: generate a full work-session package with consolidatedwork_session.txt,work_session.docx,session.ics, andsession.json; use--split-filesto also write separate helper files.scripts/make_todo.py: convert a milestone table into a todo.txt style checklist.scripts/make_ics.py: create timezone-safe calendar.icsevents for the session and its milestones, optionally with a display reminder.scripts/report_writer.py: write consolidated planning reports as TXT and DOCX without third-party dependencies.scripts/update_review_log.py: append planned-vs-actual records to a Markdown review log.scripts/reschedule_session.py: rebuild the remaining work plan after checkpoint slippage.scripts/checkpoint_session.py: record done/remaining work during execution and classify delay severity.scripts/resume_session.py: generate a resume card fromsession.jsonand local memory.scripts/handoff_session.py: create an end-of-session handoff and optionally update local memory.scripts/estimate_profile.py: summarize review logs to calibrate future time estimates.scripts/session_state.py: create, inspect, transition, and review durablesession.jsonstate.scripts/task_memory.py: append sanitized local task memory, generate a private summary, and list, forget, purge, retain, or export entries.
Output Requirements
Every plan must answer:
- What should the user do first?
- What counts as done for each milestone?
- What should be protected if time runs out?
Keep the final plan compact. Never let planning consume more value than it saves.