pm-phase-deliver
Skill1 starsChanged 42 days agoMITView on GitHub
Lean Double Diamond **Deliver phase** skill for PMs. Invoke when a feature/product is shipping or just shipped and the team needs **launch readiness, release communications, post-launch monitoring, experiment interpretation, or decision on next move** (iterate/scale/rollback/stop). Covers GTM launch readiness, release notes (user + internal + customer comms), post-launch monitoring, A/B experiment interpretation, product analytics narratives, and metric quality/guardrails. Trigger on "prepare a launch", "write release notes", "post-launch review", "did the A/B work?", "release notes para o time comercial". For funnel-shaped or growth-specific A/B tests (acquisition/activation/retention/monetisation), use `pm-archetype-growth` for the interpretation lens — this skill covers feature-level A/B and post-launch decisions across all product types.
In this repository65 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- anti-slopSkill
- data-science-analystSkill
- humanize-deliverablesSkill
- humanizerSkill
- inference-disciplineSkill
- pm-archetype-aiSkill
- pm-archetype-enterpriseSkill
- pm-archetype-growthSkill
- pm-archetype-platformSkill
- pm-phase-defineSkill
- pm-phase-deliverSkill
- pm-phase-developSkill
- pm-phase-discoverSkill
- pm-prioritization-regua-comumSkill
- pm-product-senseSkill
- pm-storytellingSkill
- pm-transversal-analysisSkill
- pm-transversal-commsSkill
- pm-transversal-docsSkill
- pm-transversal-stakeholderSkill
- repo-doctorSkill
- anti-slopSkill
- data-science-analystSkill
- humanize-deliverablesSkill
- humanizerSkill
- inference-disciplineSkill
- pm-archetype-aiSkill
- pm-archetype-enterpriseSkill
- pm-archetype-growthSkill
- pm-archetype-platformSkill
- pm-phase-defineSkill
- pm-phase-developSkill
- pm-phase-discoverSkill
- pm-prioritization-regua-comumSkill
- pm-product-senseSkill
- pm-storytellingSkill
- pm-transversal-analysisSkill
- pm-transversal-commsSkill
- pm-transversal-docsSkill
- pm-transversal-stakeholderSkill
- repo-doctorSkill
- umbertoSkill
- anti-slopSkill
- data-science-analystSkill
- humanize-deliverablesSkill
- humanizerSkill
- inference-disciplineSkill
- pm-archetype-aiSkill
- pm-archetype-enterpriseSkill
- pm-archetype-growthSkill
- pm-archetype-platformSkill
- pm-phase-defineSkill
- pm-phase-deliverSkill
- pm-phase-developSkill
- pm-phase-discoverSkill
- pm-prioritization-regua-comumSkill
- pm-product-senseSkill
- pm-storytellingSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 3 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 64
Write release communications that are fit for audience: user-facing changelog (what's new, why they'll care), internal enablement (for …
- Very specific: line 74
Run the weeks after launch as an explicit learning phase.
- Distinctive: line 94
Use funnels, retention, cohorts, and segmentation to turn behavioural data into decisions — not just awareness.
- Distinctive: line 104
Ensure metrics used for decisions are well-defined, trustworthy, and interpreted with awareness of bias, novelty effects, and side-effects.
- Distinctive: line 148
every launch has a written hypothesis and a close-out comparing actuals
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
PM Phase — Deliver
Double Diamond Phase 4 of 4. Release is the start of proof, not the end of the work.
Progressive loading
Load this SKILL.md first. For large or specialized tasks, use references/progressive-loading.md to choose the narrowest supporting reference before reading more.
When to use this skill
Invoke when:
- a launch is imminent and readiness needs to be verified
- release notes or customer/internal comms need to be drafted
- a feature has shipped and adoption/impact must be monitored
- an A/B test has finished and the ship/iterate/rollback decision is open
- a launch scorecard is due
- a post-launch review or close-out is being written
- metric quality or guardrails have thrown a flag
Trigger phrases: "prepare a launch", "write release notes", "post-launch review", "did the A/B work?", "interpret these results", "launch scorecard", "did adoption happen?", "should we ship the variant?", "release notes para o time comercial".
Don't use for discovery/strategy (→ earlier phase skills) or for post-mortem of a technical incident (that's eng's loop). For funnel-shaped or growth-specific A/B tests (acquisition / activation / retention / monetisation experiments where the discipline matters more than the launch readout), use pm-archetype-growth for the interpretation lens — this skill covers feature-level A/B and post-launch decisions across all product types.
Prime directive
A release is a hypothesis meeting reality. Expert PMs plan the learning interval as carefully as the launch itself — with a scorecard, a rollback criterion, and a close-out memo that captures what was actually learned, not what was hoped.
Calibrated disagreement
The default failure in Deliver is declaring victory on vanity metrics or conclusions the data doesn't support. Challenge premature success claims and name confounds before endorsing a launch narrative — see ../DOCTRINE.md.
Core sub-skills
1. Launch readiness and GTM coordination
Coordinate cross-functional work to bring a product/feature to market successfully and safely. A strong feature can fail commercially if readiness is weak.
Outputs: launch plan, beta plan, enablement checklist, launch FAQ, readiness sign-off, go/no-go memo.
Anti-patterns: GTM involved too late, no launch owner, missing support readiness, no launch measurement plan, "release = adoption" assumption.
→ Deep-dive: references/launch-readiness.md
2. Release notes (user + internal + customer)
Write release communications that are fit for audience: user-facing changelog (what's new, why they'll care), internal enablement (for sales/support/CS), and customer comms (email, in-app, migration notices). Release notes are the product's written face after each ship.
Outputs: public changelog entry, internal enablement one-pager, customer email, in-app release note, migration FAQ (if needed).
Anti-patterns: treating release notes as marketing, using eng-ticket language with customers, no support preparation, dumping every merged PR, vague "improvements and bug fixes" when there's a material change.
→ Deep-dive: references/release-notes.md
3. Post-launch monitoring and iteration
Run the weeks after launch as an explicit learning phase. Monitor usage, funnel, reliability, and support signals; compare actuals to hypothesis; decide iterate/scale/rollback/stop.
Outputs: launch dashboard, readout memo, close-out report, iteration backlog, rollback criteria.
Anti-patterns: launch-and-leave, measuring output volume only, no baseline, no support/rollback plan, no documented learning capture.
→ Deep-dive: references/post-launch-monitoring.md
4. Experiment design and interpretation
Interpret A/B and other product experiments with explicit hypotheses, primary metrics, guardrails, validity assumptions, and decision rules. Know when NOT to use A/B.
Outputs: experiment brief (cross-link to pm-phase-develop pre-launch), results readout, recommendation memo, holdout plan.
Anti-patterns: no primary metric, peeking without method support, underpowered tests, changing hypothesis post-hoc, shipping on noisy deltas.
→ Deep-dive: references/experiment-interpretation.md
5. Product analytics and dashboard interpretation
Use funnels, retention, cohorts, and segmentation to turn behavioural data into decisions — not just awareness.
Outputs: KPI dashboard spec, funnel analysis, retention report, cohort cut, weekly review pack.
Anti-patterns: dashboard worship, no segments, correlation-as-cause, reporting numbers without decisions.
→ Deep-dive: references/product-analytics.md
6. Metric quality and guardrails
Ensure metrics used for decisions are well-defined, trustworthy, and interpreted with awareness of bias, novelty effects, and side-effects. Design guardrails so the company does not optimise locally and harm retention, trust, or reliability.
Outputs: metric glossary, guardrail scorecard, experiment QA checklist, interpretation memo, launch-gate criteria.
Anti-patterns: one-metric obsession, shipping on averages that hide segment harm, ignoring support/trust/reliability side-effects, sloppy definitions shared across teams.
→ Deep-dive: references/metric-quality-guardrails.md
Workflow
- Load context when relevant — for project work, read
.ai/memory/active-context.mdfirst when present and resolve<slug>before opening relevant project files: PRD, tracking plan, and priorities. If no pointer exists, do not infer a project; bootstrap only when durable project context is needed and the name is known. When present,.ai/memory/org/goals.mdandcompetitors.mdframe the launch narrative. - Classify the ask — readiness, release notes, monitoring, A/B interpretation, analytics narrative, or metric-quality check?
- Anchor on the hypothesis — what were we trying to learn/achieve? State it upfront so the readout compares actuals to expectations.
- Respect the guardrails — check reliability, support load, satisfaction, and revenue side-effects before calling a result a "win".
- Decide the move — iterate, scale, hold, rollback, or stop. Decisions beat descriptions.
- Persist durable outputs — when this task creates or materially changes project context, save a launch memo, experiment result, or close-out under
.ai/memory/projects/<slug>/in the relevant files. If evidence materially changes a persona, competitor, or goal, update the matching.ai/memory/org/file and log the evidence in the relevant project changelog. A read-only answer needs no memory entry.
Output contract
## [Artefact] — [launch/experiment/release notes title]
### Hypothesis recap
### What shipped
### Primary metric + guardrails (actual vs expected)
### Segments / cohorts
### Side-effects observed
### Decision (iterate / scale / hold / rollback / stop)
### Communications (release notes / internal / customer) — linked or inlined
### Follow-up actions + owners
### Memory updatesIntegration
- Upstream:
pm-phase-develop(PRD, tracking plan, acceptance). - Transversais:
pm-transversal-docs(publishing release notes in Confluence, Jira tickets for iterate/rollback),pm-transversal-analysis(the bridge for deeper quali+quant synthesis and triangulation on user-feedback post-launch),pm-transversal-stakeholder(exec readout, DACI for ship decision). - Cross-functional pairings: your QA lead stress-tests metric interpretation and experiment integrity; your designer reviews UX post-launch; a data-science partner validates the maths when an A/B looks like a win.
- Archetype lenses (load when the product type warrants):
pm-archetype-growthfor funnel/activation/retention readouts;pm-archetype-aiwhen the launch is a model and the release gate is an eval threshold;pm-archetype-enterprisefor staged-rollout governance;pm-archetype-platformfor adoption / migration / SLO tracking.
Communication modes follow ../../docs/patterns/COMMUNICATION_MODES.md. Per-skill: Lean (default) is a 1-page hypothesis → actuals → decision → follow-up; Standard is the full launch readout with scorecard; Caveman is a 5-line shipped-X / metric-Y-vs-Z / decision / next.
Success criteria
- every launch has a written hypothesis and a close-out comparing actuals
- release notes are fit for each audience (users ≠ sales ≠ customers)
- decisions are made within a declared learning window, not by drift
- guardrail violations are caught before the feature scales
- the team accumulates a library of "what we learned" that improves the next bet
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

