project（project 效率）

Project

Why Projects Fail

Projects do not fail at the end. They fail in the middle, silently, in ways that do not become visible until it is too late to recover gracefully.

A task that was supposed to take three days has taken nine but nobody updated the tracker because updating the tracker feels like admitting failure. A dependency between two teams was identified in the planning phase, documented in a spreadsheet that nobody has opened since, and is now a blocker that will delay the launch by two weeks. The scope expanded gradually through a series of small, reasonable requests — each one taking "just a few hours" — until the project that was scoped for eight weeks is now in week fourteen with no end in sight.

None of these failures are caused by incompetent people. They are caused by the gap between how projects are planned and how projects actually unfold. Plans are static. Reality is dynamic. The gap between the two grows wider every day, and most project management tools do nothing more than document the growing distance between where you planned to be and where you actually are.

Project closes that gap. Not by creating a better plan. By creating a living system that evolves with reality, surfaces the truth about where things actually stand, and makes the corrections that keep a project moving toward completion instead of drifting toward chaos.

Starting a Project: From Ambiguity to Structure

Every project begins as a cloud of ambiguity. Someone wants something built, delivered, launched, or changed. The outcome is clear enough to describe in a sentence. The path to that outcome is anything but clear.

Describe the project. The skill does what experienced project managers do instinctively but most people skip because it takes time and discipline they do not have.

Scope definition. What is included and — critically — what is not included. The features that will be built and the features that will not. The deliverables that are promised and the deliverables that are out of scope. Written clearly enough that when the client asks "can you also add X?" six weeks in, there is a documented reference point for whether X was always part of the plan or is a scope change that requires a timeline adjustment.

Phase breakdown. The project divided into sequential phases, each with its own objective, deliverables, and completion criteria. Not a flat list of tasks. A structure that shows which work enables which other work, where the critical path runs, and which phases can overlap.

Task decomposition. Each phase broken into specific tasks small enough to be assigned to one person and completed in one to five days. Tasks that are larger than five days are hiding complexity that will surprise you later. The skill identifies these and breaks them further until every task is concrete, measurable, and assignable.

Dependency mapping. Which tasks cannot start until other tasks finish. Which teams need to hand off work to other teams. Which external inputs — client approvals, vendor deliveries, regulatory reviews — gate progress on internal work. Dependencies are where projects die. Making them visible is how projects survive.

Timeline construction. Working backward from the deadline or forward from today, the skill builds a realistic timeline that accounts for dependencies, resource availability, and the historical reality that humans underestimate task duration by 30 to 50 percent. The timeline includes buffer — not as padding but as a mathematical acknowledgment that not everything will go according to plan.

Risk identification. What could go wrong. Not every theoretical risk — that list is infinite and useless. The three to five specific risks that, based on the project type and team composition, are most likely to materialize. For each risk: the probability, the impact, the early warning signs, and the mitigation plan.

Tracking: The Truth About Where Things Stand

A project plan is a hypothesis. Tracking is the experiment that tests it.

Most project tracking fails because it relies on humans voluntarily updating their status in a tool they do not want to open. The result is a project tracker that shows what people remember to update rather than what is actually happening. The dashboard says 73% complete. Reality says something different but nobody knows exactly what because the dashboard has not been accurate since week two.

The skill tracks projects through conversation rather than forms. When you mention in a chat that the design mockups are done, the skill updates the tracker. When a team member says they are blocked waiting on API access, the skill logs the blocker, identifies who can resolve it, and flags it in the next status report. When a deadline slips, the skill recalculates the downstream impact immediately rather than waiting for someone to manually adjust twenty linked dates in a Gantt chart.

Daily pulse. A thirty-second summary of where the project stands right now. Tasks completed yesterday. Tasks in progress today. Blockers that need attention. Deadlines approaching in the next five days. No dashboard required. The project status comes to you.

Blocker escalation. A task has been blocked for three days. The person who can unblock it has not responded to two messages. The skill escalates — not by sending an angry email but by surfacing the blocker to the right person with full context: what is blocked, why it matters, what the downstream impact will be if it is not resolved by a specific date.

Velocity tracking. How fast is work actually getting done compared to how fast it was planned to get done? If the team is completing tasks at 60% of the planned rate, the skill recalculates the projected completion date immediately. You see the real timeline, not the aspirational one.

Scope Management: The Art of Saying Not Yet

Scope creep is not caused by malice. It is caused by the natural human tendency to improve things. Someone has a good idea. The idea is genuinely good. It would make the project better. So it gets added. Then another idea. Then another. Each one small. Each one reasonable. Together, they transform a focused project into an expanding universe of requirements that no timeline can contain.

The skill maintains a clear boundary between what is in scope and what is not. When a new request arrives — from the client, from the team, from your own inspiration at 2am — the skill evaluates it against the current plan.

If the request fits within existing scope, it is absorbed. If it does not, the skill frames the tradeoff clearly: adding this feature will extend the timeline by approximately X days, or it can replace this other feature of similar size, or it can be added to a phase two backlog. The decision is yours. The clarity about the consequence of each decision is the skill's job.

This is not about saying no. It is about saying yes with full understanding of what yes costs.

Status Reports: From Fiction to Data

Friday afternoon. The status report is due. You open a blank document and try to reconstruct what happened this week. You remember the big things — the milestone that was hit, the blocker that was resolved. You forget the small things that actually consumed most of the time. You phrase everything optimistically because nobody wants to write a status report that says "we are behind and I am not sure why."

The skill generates status reports from actual project data. Tasks completed this week with dates and owners. Tasks that slipped and why. Blockers resolved and blockers still open. Budget consumed versus budget remaining. Timeline status: on track, at risk, or behind, with specific evidence for the assessment.

The report is honest because it is generated from data rather than memory. It is useful because it highlights deviations from the plan rather than reciting activities. And it takes zero effort because the tracking happened continuously throughout the week rather than in a Friday afternoon reconstruction sprint.

For different audiences, the same underlying data produces different reports. The client gets a high-level summary focused on deliverables and timeline. The team gets a detailed operational view focused on next week's priorities. Leadership gets a dashboard focused on budget, risk, and strategic milestones. One set of data. Three views. None of them requiring separate writing.

Multi-Project Management

Most professionals do not manage one project. They manage three, or five, or twelve, each at a different stage, each with its own stakeholders, timelines, and urgency levels.

The skill maintains a portfolio view across all active projects. Which projects are on track and can run on autopilot this week. Which projects have issues that need your attention today. Which projects are approaching milestones that require your involvement.

Resource conflicts — when the same person is assigned critical tasks on two projects with overlapping deadlines — are surfaced before they become crises. When your designer is the bottleneck on both the website redesign and the product launch materials, the skill identifies this and presents options: delay one project, bring in additional help, or reduce scope on one of the deliverables.

Project Closure: Ending With Intention

Most projects do not end. They fade. The last few tasks linger incomplete for weeks. The final deliverable is sent but the retrospective never happens. The lessons learned are lost because nobody took the time to capture them while the experience was fresh.

The skill ensures clean closure. Final deliverable checklist: everything that was promised, verified as delivered. Outstanding items documented with clear ownership and deadlines. Budget reconciliation: what was planned versus what was spent and why the difference exists.

Retrospective prompts based on the specific project: what went well, what went wrong, what took longer than expected and why, what would you do differently. These are captured and stored as institutional knowledge that makes the next project better. Not in a document that nobody reads. In the skill's memory, where it will surface relevant lessons when you start a similar project in the future.

Who This Is For

Freelancers and solopreneurs managing multiple client projects with no project manager, no Gantt chart, and no time to learn enterprise project management tools.

Team leads who inherited a project that was already behind schedule and need to understand the real status before they can fix it.

Product managers coordinating across engineering, design, marketing, and leadership to ship something on a date that everyone agreed to and nobody believes.

Agency owners running ten client projects simultaneously and needing every one of them to deliver on time because reputation is the only asset that matters.

Anyone who has ever looked at a project plan on day one and thought "this seems reasonable" and then looked at reality on day sixty and wondered what happened.

Privacy

All project data stays within your agent memory. Client information, internal timelines, budget details, team performance data — none of it is shared externally. Status reports are generated for your review before distribution. Nothing leaves your control without your explicit approval.