docs(agents): add Fail Loud, Never Fake error-handling rule#101
Conversation
Adds a new subsection under Coding Standards & Principles establishing that visible failure is preferred over silent fallbacks. Codifies a priority ladder (works → disclosed fallback → clear error → never silent), forbidden patterns, and agentic directives forbidding swallowed exceptions, fake/placeholder data substitution, and false success reports. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
There was a problem hiding this comment.
Pull request overview
Adds a new documentation rule to the agent/human contribution guidelines to prevent silent fallbacks and “fake success” outcomes, complementing the existing defensive coding and fail-fast principles.
Changes:
- Introduces a “Fail Loud, Never Fake” subsection with a clear priority ladder for handling failures.
- Documents explicitly forbidden error-handling patterns (swallowed exceptions, synthetic substitutions, false success reporting, test weakening).
- Clarifies that disclosed fallbacks are acceptable when observability is maintained (logs/flags/metrics/UI).
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: ASSERTIVE Plan: Pro Run ID: 📒 Files selected for processing (1)
📝 WalkthroughWalkthroughAdds a "Fail Loud, Never Fake" section to Changes
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~3 minutes Possibly related PRs
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Two new lines exceeded the 200-char markdownlint limit. Wrap them at sentence boundaries to match the style used in the surrounding sections. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
|
@coderabbitai full review |
✅ Actions performedFull review triggered. |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Automated review — APPROVEDRisk: LOW SummaryDocumentation-only change adding a 'Fail Loud, Never Fake' error-handling guideline to AGENTS.md. All CI checks pass (Lint, ShellCheck, CodeQL, SonarCloud, AgentShield), SonarCloud reports zero issues and zero security hotspots, and CodeRabbit has already approved. The content is well-structured, internally consistent, and complementary to the existing 'Fail fast' and defensive-coding sections with no security, correctness, or maintainability concerns. FindingsInfo
CI statusAll status checks passed: Lint, ShellCheck, AgentShield, Agent Security Scan, CodeQL, SonarCloud (0 new issues, 0 security hotspots), CodeRabbit (APPROVED). No failures or warnings. Reviewed by the don-petry PR-review cascade (triage: haiku 4.5 → deep: sonnet 4.6 + duck: gpt-5.4 → audit: opus 4.6). Reply with |
Automated review — APPROVEDRisk: LOW
SummaryDocumentation-only PR adding a 'Fail Loud, Never Fake' error-handling section to AGENTS.md. Triage escalated due to a triage-tier crash (triage-output-invalid), not any actual risk signal — the prior review at f247b89 already confirmed this is LOW risk. All CI checks pass (Lint, ShellCheck, AgentShield, Agent Security Scan, CodeQL, SonarCloud with 0 issues/0 hotspots, CodeRabbit APPROVED), no code or config files modified, no new dependencies. FindingsInfo
CI statusAll CI checks passed at head SHA Reviewed by the don-petry PR-review cascade (triage: haiku 4.5 → deep: sonnet 4.6 + duck: gpt-5.4 → audit: opus 4.6). Reply with |
Automated review — APPROVEDRisk: LOW SummaryDocumentation-only PR adding a 'Fail Loud, Never Fake' error-handling section to AGENTS.md (+32/-0). Triage escalated due to a triage-tier crash (triage-output-invalid), not any genuine risk signal. All CI checks pass (Lint, ShellCheck, AgentShield, Agent Security Scan, CodeQL, SonarCloud with 0 new issues and 0 security hotspots, CodeRabbit APPROVED), no code/config/secrets/CI-workflow files were modified, and no new dependencies introduced. FindingsInfo
CI statusAll status checks passed at head SHA Reviewed by the don-petry PR-review cascade (triage: haiku 4.5 → deep: sonnet 4.6 + duck: gpt-5.4 → audit: opus 4.6). Reply with |
|
|
* docs(agents): add Fail Loud, Never Fake error-handling rule Adds a new subsection under Coding Standards & Principles establishing that visible failure is preferred over silent fallbacks. Codifies a priority ladder (works → disclosed fallback → clear error → never silent), forbidden patterns, and agentic directives forbidding swallowed exceptions, fake/placeholder data substitution, and false success reports. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * docs(agents): wrap long lines to satisfy MD013 (max 200) Two new lines exceeded the 200-char markdownlint limit. Wrap them at sentence boundaries to match the style used in the surrounding sections. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> --------- Co-authored-by: don-petry <don-petry@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* docs(agents): add Fail Loud, Never Fake error-handling rule Adds a new subsection under Coding Standards & Principles establishing that visible failure is preferred over silent fallbacks. Codifies a priority ladder (works → disclosed fallback → clear error → never silent), forbidden patterns, and agentic directives forbidding swallowed exceptions, fake/placeholder data substitution, and false success reports. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * docs(agents): wrap long lines to satisfy MD013 (max 200) Two new lines exceeded the 200-char markdownlint limit. Wrap them at sentence boundaries to match the style used in the surrounding sections. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> --------- Co-authored-by: don-petry <don-petry@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* docs(agents): add Fail Loud, Never Fake error-handling rule Adds a new subsection under Coding Standards & Principles establishing that visible failure is preferred over silent fallbacks. Codifies a priority ladder (works → disclosed fallback → clear error → never silent), forbidden patterns, and agentic directives forbidding swallowed exceptions, fake/placeholder data substitution, and false success reports. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * docs(agents): wrap long lines to satisfy MD013 (max 200) Two new lines exceeded the 200-char markdownlint limit. Wrap them at sentence boundaries to match the style used in the surrounding sections. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> --------- Co-authored-by: don-petry <don-petry@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* docs(agents): add Fail Loud, Never Fake error-handling rule Adds a new subsection under Coding Standards & Principles establishing that visible failure is preferred over silent fallbacks. Codifies a priority ladder (works → disclosed fallback → clear error → never silent), forbidden patterns, and agentic directives forbidding swallowed exceptions, fake/placeholder data substitution, and false success reports. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * docs(agents): wrap long lines to satisfy MD013 (max 200) Two new lines exceeded the 200-char markdownlint limit. Wrap them at sentence boundaries to match the style used in the surrounding sections. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> --------- Co-authored-by: don-petry <don-petry@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Summary
Adds a new Fail Loud, Never Fake subsection to
AGENTS.mdunder Coding Standards & Principles, immediately after Defensive Coding at System Boundaries. The new rule codifies a long-standing failure mode: agents (and humans) hiding errors behind silent fallbacks, placeholder data, or false success reports — turning what should be a 5-minute stack-trace fix into a day lost chasing wrong data downstream.The section establishes:
except:swallowing, synthetic data substitution, false "I've set up X" reports, weakening tests to make them pass)Why
This complements the existing Fail fast bullet, which covers invariant violations, by addressing the adjacent (and more insidious) problem of work that appears to succeed but didn't. The two rules reinforce each other.
Test plan
.markdownlint-cli2.yaml)Summary by CodeRabbit