prd-delivery-lifecycle
Skill1 starsChanged 3 months agoMITView on GitHub
Drive a product requirement from evidence-backed definition through implementation, verification, acceptance, and release readiness. Use for multi-step feature delivery, PRD authoring, requirement changes, feedback rounds, traceability, or cross-component work where decisions and evidence must remain coherent.
In this repository10 other files it gives its agents.
- auto-token-trade-safety-gateSkill
- binance-major-crypto-trendSkill
- codex-session-retrospectiveSkill
- durable-task-handoffSkill
- flutter-flow-auditSkill
- new-token-risk-analysisSkill
- polish-uiSkill
- release-sql-gateSkill
- ui-expertSkill
- vrf-consumer-auditSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 7 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 14
Read evidence-contract.md for evidence layers and state-machine.md for feedback and invalidation rules.
- Distinctive: line 30
Keep durable identifiers and links for:
- Distinctive: line 36
Every implementation slice must trace back to a requirement and forward to verification.
- Distinctive: line 43
add or identify the failing test or reproducible evidence.
- Distinctive: line 45
run targeted checks, then broader checks proportional to risk.
- Distinctive: line 46
update trace links, evidence, and unresolved risks.
- Distinctive: line 49
Do not equate static analysis with runtime behavior, a passing test with deployment, or deployment with user acceptance.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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
PRD Delivery Lifecycle
Maintain one versioned delivery model across three connected loops:
- Define and authorize — establish the problem, scope, decisions, contracts, risks, and acceptance evidence.
- Build and prove — implement vertical slices with tests and layer-appropriate validation.
- Accept and prepare release — reconcile feedback, close evidence gaps, and identify remaining authority gates.
Read evidence-contract.md for evidence layers and state-machine.md for feedback and invalidation rules.
Start from authoritative evidence
Before editing:
- identify the requirement owner and authoritative source;
- separate facts, assumptions, policy choices, and unresolved questions;
- enumerate producers, consumers, states, failure paths, and compatibility surfaces;
- define explicit non-goals and authority boundaries;
- state which evidence will prove each acceptance criterion.
Reuse the repository's existing product-spec location. If none exists, create the smallest coherent feature record containing source, review, decisions, implementation plan, verification, and release notes.
Maintain the trace graph
Keep durable identifiers and links for:
REQ -> DEC -> CONTRACT -> TEST -> SLICE -> VERIFY -> RELEASEEvery implementation slice must trace back to a requirement and forward to verification. Every changed decision must identify which downstream nodes are stale.
Execute in vertical slices
For each slice:
- choose one user-observable or contract-observable outcome;
- add or identify the failing test or reproducible evidence;
- implement the smallest coherent change;
- run targeted checks, then broader checks proportional to risk;
- update trace links, evidence, and unresolved risks;
- request acceptance only when the required evidence layer exists.
Do not equate static analysis with runtime behavior, a passing test with deployment, or deployment with user acceptance.
Process feedback
Convert each feedback item into a packet containing the claim, evidence, impacted identifiers, classification, decision owner, and disposition. Invalidate only the affected downstream graph, then rerun the required checks. Never silently rewrite the requirement history.
Completion contract
Report:
- behavior before and after;
- requirement-to-evidence trace status;
- changed components and their responsibilities;
- validation by layer and its limitations;
- unresolved decisions, risks, migrations, rollout, and rollback needs;
- exact authority still required for commit, push, merge, deploy, or release.
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.

