PromptBase
PromptBase SelectGet Select
Close icon
General
Home
Marketplace
Create
Hire
Login
Chat
Sell
Explore

Prompt Details

Model
Chat - (gpt-5.6)
Token size
488
Example input
[INCIDENT BRIEF]: Some users report intermittent login failures. Expected successful login; actual occasional 500 responses. Reproduction and recent changes unknown. [PASTE ERRORS OR LOGS]: Only HTTP 500, no stack trace, request ID or server logs. [PASTE CODE OR FILES]: None supplied. [STACK / VERSION / DEPLOYMENT DETAILS]: Web service; stack and version unknown.
Example output
1. Summary of Issue Observed facts Some login attempts intermittently return HTTP 500. Expected behavior: Plain text Successful login Actual behavior: Plain text Occasional HTTP 500 Known evidence is limited to: Plain text HTTP 500 There is: no stack trace no request ID no server log no relevant code no known reproduction pattern no known recent change no confirmed application stack Because of that, the root cause is not established from the supplied evidence. A 500 only proves that the server encountered an unhandled or internally classified failure somewhere in the login request path. It does not identify whether the failure is in authentication logic, database access, session creation, an external identity provider, infrastructure, or another dependency. 2. Root Cause Verified cause Not yet determined. There is insufficient evidence to identify a primary cause without guessing. Smallest diagnostic needed The highest-value next diagnostic is: Capture one failing login request with its corresponding server-side exception/error log and correlation/request identifier. Specifically, for one 500, obtain: Plain text timestamp request path request ID / correlation ID exception type stack trace upstream/dependency error, if any HTTP status returned That single artifact will usually separate application-code failures from dependency or infrastructure failures. Current hypotheses These are hypotheses only, not confirmed causes. Because the failure is intermittent, plausible categories include: Database or connection-pool failure transient connection exhaustion timeout failed user/session lookup deadlock or database unavailability Session/token creation failure cache/session-store outage signing/key access issue serialization failure External authentication dependency OAuth/OIDC/LDAP/identity-provider timeout rate limiting malformed or unexpected upstream response Unhandled input-dependent exception only certain users or account states trigger the error null/invalid profile or credential-related field Race or concurrency issue duplicate session creation concurrent login/update conflict Infrastructure instability one unhealthy application instance load balancer routing to a broken node memory/resource exhaustion The evidence does not currently distinguish among these. Related components The login path likely involves some subset of: Plain text HTTP route/controller authentication service user repository/database password/credential verifier session or token service cache/session store external identity provider middleware reverse proxy/load balancer Which of these is actually affected remains unknown. 3. Impact Assessment Application impact Confirmed impact: some users cannot authenticate successfully failed requests return a generic 500 The intermittent nature means the service is not completely unavailable. The percentage of affected users and requests is unknown. Files affected No exact files can responsibly be named because no source tree or stack has been supplied. Any claim such as: Plain text auth.js login.py session_service.go would be invented. The specific file should only be identified after the failing stack trace points to the relevant code path. Database impact Unknown. There is no evidence of: corruption failed writes schema problems transaction failures Database involvement is a plausible hypothesis because authentication commonly queries user state, but it is not established. API impact Confirmed: Plain text login endpoint occasionally returns HTTP 500 Potential downstream effects include: failed client authentication retry loops degraded availability elevated error rate No broader API impact is established. Security implications There are two relevant concerns. First, authentication failures on their own do not prove a security vulnerability. Second, generic 500 responses are preferable to returning raw stack traces or sensitive authentication details to clients. However, server-side diagnostics must still retain sufficient detail for investigation. Additional security-relevant hypotheses include: account-specific malformed state token/signing failures identity-provider issues abuse or rate-limit edge cases None are currently verified. 4. Recommended Fix A code fix should not be applied yet because the failure mechanism is unknown. Immediate minimal action Instrument or retrieve diagnostics around the login request path. At minimum, each login request should emit a correlation ID and structured error record containing: Plain text request_id timestamp route application instance authentication stage exception class exception message dependency name latency final status Sensitive values must not be logged, including: Plain text passwords session tokens access tokens refresh tokens credential hashes full authorization headers Minimal investigation sequence For the next failing request: Plain text 1. Identify the exact request timestamp/request ID. 2. Locate the server-side exception. 3. Determine the failing stage: route -> user lookup -> credential verification -> session/token -> response. 4. Check dependency status at that timestamp. 5. Compare with a successful request for the same path. If there are multiple application instances, also record the instance/container/host. Intermittent failures caused by one unhealthy node often become immediately visible this way. Smallest safe fix The smallest safe fix depends on the evidence. Examples: If logs show: Plain text database connection timeout the minimal fix may involve database/pool handling. If logs show: Plain text null reference in session creation the change should be confined to that specific session path. If all failures originate from one application node: Plain text instance-3 the immediate remediation may be operational rather than a source-code change. Until one of those is established, changing authentication code would violate the minimal-change requirement and risks masking the real issue. 5. Testing Strategy Because the root cause is still unknown, testing should first focus on reproducing and isolating the failure. Diagnostic reproduction matrix Test login attempts across: Plain text same user repeatedly multiple affected users known-good user correct credentials incorrect credentials new session existing session different application instances, if visible Record whether failures correlate with: Plain text specific users specific nodes specific timestamps request volume database latency external identity-provider latency session-store state Regression testing Once the actual cause is found, add a regression test reproducing that exact failure condition. For example, if the cause turns out to be a transient dependency timeout, the test should simulate that dependency condition and verify the intended behavior. If the cause is malformed user state, create a fixture matching that state. Do not create speculative tests for unrelated causes before the failure mechanism is known. Monitoring improvements For production, add or confirm: Plain text login success rate login 5xx rate 5xx count by application instance authentication latency database/pool errors session-store errors external identity-provider errors Alerting should distinguish: Plain text authentication rejection -> expected 4xx internal authentication failure -> unexpected 5xx Verification criteria A fix is only verified when: Plain text the original failing condition can be reproduced before the fix, the specific failure no longer occurs after the fix, normal successful login still works, expected authentication failures still return the correct non-500 status, and no sensitive information is exposed in responses or logs. No tests have been executed here. At this point, the correct engineering conclusion is: the incident is confirmed, but the root cause is not yet diagnosable from the available evidence. The next required artifact is one correlated server-side error/stack trace from a failing login request.
GPT

Bug Triage Minimalfix Analysis

Add to Cart
PromptBase Select
Get this prompt + 9 more with Select$14 your first month, then $19/mo$14 first month, then $19/mo
View Select
Instant accessInstant access
Usage rightsCommercial use
Money-back guaranteeMoney‑back
By purchasing this prompt, you agree to our terms of service
GPT-5.6
Tested icon
Guide icon
4 examples icon
Free credits icon
Investigate software defects using supplied code, logs, observed behaviour and environment details. Get an issue summary, evidence-based root-cause assessment, impact analysis, a minimal correction and regression-test strategy. Separates hypotheses from findings and requests missing evidence instead of inventing a diagnosis. Includes four worked debugging examples covering JavaScript, Python and incomplete incident reports.
...more
Added 2 weeks ago
Report
Browse Marketplace