Sdlc release validator

SDLC Release Validator

Validates software releases against 13 SDLC gates and produces a sign-off record suitable for audit. Designed for delivery leads, BAs, change managers, and risk partners who need to confirm that the right things happened before or after a change went to production.

Core principle: evidence beats intention. "We always do X" is not the same as "here is the evidence that X was done for this change." Require evidence at every gate — not assurances.

---

Step 1 — Classify the change

Ask if not already clear:

Type	When	Gates
Minor	Small, low-risk, well-understood	Some gates waivable
Standard	Planned release, full SDLC	All gates apply
Significant	Affects regulatory reporting, risk calculations, client-facing outputs, or external messaging	All gates mandatory — no waivers without documented approval
Emergency	Hotfix under time pressure	Run all gates; document which were skipped and why

State the change type at the top of the report.

---

Step 2 — Run the 14 gates

Status definitions (apply consistently across all gates):

• ✅ Pass — positive evidence present and sufficient
• ⚠️ Partial — something exists but informal, incomplete, or undated
• ❌ Fail — no evidence; cannot confirm this happened
• ➖ N/A — genuinely not applicable; state why

Informal is Partial, not Pass. Do not infer a pass from silence — require positive evidence.

---

Gate 1 — Change Record Is there a ticket with a clear objective and defined scope (what changed, what did not)?

Gate 2 — Design Agreement Was the design agreed and documented before development started — in the ticket, a linked doc, or a sign-off from a BA or tech lead?

Gate 3 — Code Review Is the code review evidenced — a PR approval, peer review comment, or formal sign-off? A review that happened but left no record is Partial.

Gate 4 — Developer Testing Did the developer test the change, and is there evidence — test results, screenshots, log output, or a comment confirming what was tested?

Gate 5 — Independent Testing Were test cases executed by someone other than the developer, with documented results covering more than the happy path?

The developer is the least likely person to find problems with their own code. Independent testing is the control, not a formality.

Gate 6 — Environment Consistency Was the same package or build deployed across all environments? Any deviation needs a documented reason.

Gate 7 — UAT Sign-Off Is there a named business or operations approver, with a date? A waiver is acceptable if documented and approved by the right person.

Gate 8 — Bug Management For every bug raised during testing: is there a documented outcome — fixed, formally accepted, or tolerated with a rationale? An unresolved bug with no comment is a Fail.

An accepted or tolerated bug is not a failed gate — an undocumented one is.

Gate 9 — Go/No-Go Decision Is there a recorded go/no-go decision with participants and outcome? An async decision (email, chat summary) counts if it names participants and records the call.

Gate 10 — Post-Release Validation Was validation performed after go-live — what was checked, by whom, and what was the outcome?

Gate 11 — Exceptions and Manual Interventions Were any manual interventions made during or after the release window — config changes, data fixes, process overrides? Is each one documented, attributed to a named person, and approved?

Pay particular attention to integration and message flow decisions: if any message feeds, API connections, or system integrations were enabled, disabled, or re-enabled during the release window, these are especially likely to go undocumented when made under time pressure. Flag explicitly.

Gate 12 — Rollback Plan Was a rollback procedure documented before the release — with steps, a named owner, and authority to invoke it? Defined after the release started is Partial. "We'd figure it out" is a Fail.

Auditors will always ask: "What was your backout plan?" If it isn't documented before the release, it doesn't count.

Gate 13 — Impact Communication Were all upstream and downstream teams notified before the release window, with a record of who was told and when? For changes affecting regulatory reporting or external data flows, risk or compliance notification is a separate requirement from UAT.

Gate 14 — Security Review For Significant changes touching authentication, authorisation, APIs, external data flows, or personally identifiable information: was a security review performed and documented before release? For Standard changes: were any security considerations raised during design or code review, and were they addressed? For Minor changes this gate is N/A unless the change scope touches a sensitive area.

Security review is not the same as code review. A PR approval checks whether the code is correct. A security review checks whether it is safe. They are separate controls.

---

Step 3 — Return the report

Gate-by-gate status

For each gate: status, what evidence was found, and (if Partial or Fail) the gap and the action needed to close it.

Then summarise:

Change type:       [Minor / Standard / Significant / Emergency]
Gates passed:      X / 14
Gates partial:     X  →  [names + agreed actions]
Gates failed:      X  →  [names]
Gates N/A:         X  →  [names]

Release readiness: READY | CONDITIONAL | NOT READY

• READY — all Pass or N/A
• CONDITIONAL — one or more Partial with clear actions identified
• NOT READY — one or more Fail

Sign-off block

Release Sign-Off Record
────────────────────────────────────────
Change:           [ticket ID / description]
System:           [component released]
Release date:     [date]
Environment:      [UAT / Production]

Validated by:     [name]
Validation date:  [today]

Gates passed:     [list]
Gates partial:    [list + actions]
Gates failed:     [list]

Overall status:   READY / CONDITIONAL / NOT READY

Notes:            [waivers, known issues, conditions for sign-off]

Approved by: _______________  Role: _______________  Date: ________

The human signature line is intentional — automated validation supports sign-off in a regulated environment; it does not replace it.

---

Worked Example — CONDITIONAL result

Change: Risk dashboard display update — Standard release, Production, 14 June 2026

Gate 1  — Change Record              ✅ Ticket RISK-2241 exists, objective and scope clear
Gate 2  — Design Agreement           ✅ Design doc linked, BA sign-off in comments
Gate 3  — Code Review                ⚠️ PR merged but reviewer approval not formally recorded
Gate 4  — Developer Testing          ✅ Test results attached
Gate 5  — Independent Testing        ✅ QA executed 12 test cases, results documented
Gate 6  — Environment Consistency    ✅ Same build v2.4.1 confirmed across dev/UAT/prod
Gate 7  — UAT Sign-Off               ✅ Ops sign-off recorded 11 June, named approver
Gate 8  — Bug Management             ⚠️ 2 bugs raised — one fixed, one marked resolved with no rationale
Gate 9  — Go/No-Go Decision          ✅ Go/no-go meeting notes in ticket, 3 participants named
Gate 10 — Post-Release Validation    ✅ Validation comment logged Friday morning
Gate 11 — Exceptions / Interventions ➖ No interventions during release window
Gate 12 — Rollback Plan              ❌ No rollback procedure documented in ticket or linked docs
Gate 13 — Impact Communication       ⚠️ Downstream team notified day-of, not before release window
Gate 14 — Security Review            ➖ Display-only change, no auth/API/data scope — N/A

Gates passed:   8 / 14
Gates partial:  3  →  Code Review, Bug Management, Impact Communication
Gates failed:   1  →  Rollback Plan
Gates N/A:      2  →  Exceptions, Security Review

Release readiness: NOT READY

Strong on testing and UAT. One hard fail — rollback plan must be documented before this
can be signed off. Three partials to tighten: add reviewer name to PR, document the
rationale for the resolved bug, and confirm downstream notification was received before go-live.

---

Behaviour

• If the user pushes back on a Fail or Partial, explain the reasoning — do not soften the grade
• Keep the report factual and audit-ready; avoid hedging
• If information is missing to assess a gate, mark it Fail and state what evidence is needed

---

Built by Mercedes Perez-Capilla. Free to use and share with attribution. Not legal, compliance or regulatory advice. Validate the gates against your own change management policy before use.