COPR ACP cheat sheet

Agile Manifesto: four value pairs favor individuals/interactions, working software, customer collaboration, and responding to change over their right-hand counterparts.

Scrum: three roles (Product Owner, Scrum Master, Developers), five events, three artifacts; sprints are timeboxed to one month or less with a fixed goal.

Kanban core practices: visualize work, limit WIP, manage flow, make policies explicit, use feedback loops, improve collaboratively.

Little's Law: average lead time = average WIP / average throughput; cutting WIP or raising throughput shortens delivery time.

Lean seven wastes (TIMWOOD): Transport, Inventory, Motion, Waiting, Overproduction, Overprocessing, Defects.

Cumulative flow diagram: widening vertical band = growing WIP; flat horizontal band = blocked flow; slope of a band = throughput.

XP practices include pair programming, test-driven development, continuous integration, small releases, and collective code ownership.

Definition of Done is a shared, explicit checklist applied to every increment; incomplete items return to the backlog.

Prioritize by value density: value delivered divided by effort or cost of delay; highest ratio first maximizes early return.

Cost of Delay is the money lost per unit time a feature is late; WSJF = Cost of Delay / Job Size (or Duration).

MoSCoW categories: Must have, Should have, Could have, Won't have this time; Musts must be independently deliverable.

Minimum Marketable Feature / MVP is the smallest release that delivers value or learning; scope is trimmed, not quality.

Value stream mapping distinguishes process time from lead time; flow efficiency = value-add time / total lead time.

Kano model classes: Must-be, Performance, Attractive, Indifferent, Reverse; only performance and attractive features differentiate.

Story splitting patterns: workflow steps, business rules, data variations, interfaces, and happy path versus edge cases.

Real options reasoning: defer, stage, or abandon investment; option value rises with uncertainty and falls with irreversibility.

Power/interest grid: high power-high interest = manage closely; high power-low interest = keep satisfied; low power-high interest = keep informed.

Stakeholder analysis records influence, interest, attitude (supporter, neutral, blocker), and preferred communication channel.

Communication plan fields: audience, message, channel, frequency, owner, and feedback mechanism.

Product Owner is the single voice for backlog ordering; stakeholders route requests through the PO, not directly to the team.

Reviews/demos are the primary feedback loop; invite real users to inspect a working increment, not a status report.

Onion model of stakeholders: team, direct users, indirect users, and wider organization; nearer rings inform design first.

Escalation path is defined in advance: team, then Scrum Master/coach, then product leadership, with time triggers.

RACI confusion (many accountable) causes decision stalls; exactly one accountable owner per decision.

Tuckman stages: Forming, Storming, Norming, Performing, Adjourning; conflict peaks in Storming and is normal.

Servant leader behaviors: remove impediments, shield from interruption, coach, facilitate decisions, and let the team own technical choices.

Team size guidance for Scrum: small enough to coordinate, typically 3-9 members; larger groups split into multiple teams.

Self-organizing teams select how much work to take on; the leader sets constraints, not assignments.

Conflict modes (Thomas-Kilmann): competing, collaborating, compromising, avoiding, accommodating; collaborating suits shared goals.

Psychological safety predicts learning behavior; leaders model fallibility and reward surfacing problems.

Team charter/working agreements define definition of done, meeting norms, and decision rules before conflict arises.

Velocity is a planning aid for one team only; it is not a productivity metric for cross-team comparison.

Rolling-wave planning: detailed near term, coarse long term; re-plan each iteration using new information.

Release planning uses velocity range (best/average/worst) to forecast dates, not a single-point commitment.

Story points measure relative size including complexity, risk, and unknowns; they are unitless and team-specific.

Ideal days versus story points: never mix the two within one backlog; convert only for rough external forecasts.

Planning poker: simultaneous reveal reduces anchoring; re-vote after discussion; outliers explain reasoning.

Timeboxed ceremonies: daily 15 minutes, sprint planning and review/retrospective proportional to sprint length.

Buffer techniques: feature buffer, schedule buffer, and capacity buffer absorb uncertainty without inflating every estimate.

Backlog refinement keeps the top of the backlog ready; a common rule is enough refined work for two iterations.

Impediment log: record blocker, owner, date raised, and resolution; daily standup surfaces new ones.

Cycle time versus lead time: cycle time starts at work start, lead time at request; both come from board timestamps.

Control chart (run chart) rules: seven consecutive points on one side of the mean signals a non-random shift.

Root-cause tools: five whys, fishbone (Ishikawa), and fault tree; fix causes, not symptoms.

Escaped defects and reopened items are lagging indicators; WIP age and blocked time are leading indicators.

Blocker escalation rule: if unresolved past a set threshold (for example one day), escalate to the next level.

Anti-pattern: adding WIP to speed up delivery lengthens lead time per Little's Law.

Variance analysis compares planned versus actual at iteration close; unexplained variance triggers a retrospective topic.

Retrospective structure: set the stage, gather data, generate insights, decide actions, close; limit actions to a few owned items.

Kaizen: small continuous improvements by the people doing the work; PDCA (Plan-Do-Check-Act) is the cycle.

Technical excellence practices: refactoring, automated tests, continuous integration, and coding standards protect velocity.

Test pyramid: many unit tests, fewer service/integration tests, few end-to-end UI tests; invert only with justification.

Definition of Done gaps appear as technical debt; track debt explicitly and budget capacity to repay it.

Quality is built in, not inspected in: acceptance criteria are agreed before development begins.

Metrics for improvement: escaped defects, cycle time, throughput, and change failure rate; avoid individual utilization targets.

Improvement backlog items are prioritized like product work and reviewed for effect at the next retrospective.

Risk exposure = probability x impact; rank risks and re-score at each iteration or release boundary.

Risk response types: avoid, mitigate, transfer, accept, and exploit (for opportunities).

Spike is a timeboxed investigation to reduce technical or requirements uncertainty before committing.

ROAM board for risks: Resolved, Owned, Accepted, Mitigated; each risk gets one classification and an owner.

Regulatory constraints (audit trails, data privacy, accessibility) become acceptance criteria and Definition of Done items, not afterthoughts.

Compliance evidence is produced continuously by automated pipelines and traceable backlog items rather than a final documentation phase.

Vendor and dependency risk: track external lead times and single points of failure on the dependency board.

Architectural runway: intentional up-front design sufficient for near-term features; too little causes rework, too much is waste.