Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsBlogPro
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Authors
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges
  • Chrome Extension
  • Skill Manager

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

ProTermsPrivacyRefunds
Back to skills

Debug Incident Protocol

ASecurity

Use when the user says: «it doesn''t work», «still broken», «hung/stuck», «disappeared after the update», «it used to work», «the metric is zero but the UI is fine» — or when an incident needs to be analyzed, the root cause of a hang found, and whether a process was actually restarted verified. Covers: facts before theories (storage/logs/PID), symptom vs root cause, silent failure, restart ritual, localizing hangs with a progress marker, timeouts, cache masking. Do not use for writing tests (...

3 stars
0 votes
0 copies
0 views
Added 9/22/2026
ai-agentstestinggit

Works with

cli

Security Analysis

A100/100

Scanned 10/6/2026

$npx -y skills add oleg494/coding-kit --skill debug-incident-protocol --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Debug Incident Protocol?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Debug Incident Protocol
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/oleg494-debug-incident-protocol/badge)](https://www.skillsdirectory.com/skills/oleg494-debug-incident-protocol)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
Files
SKILL.md
---
name: debug-incident-protocol
description: 'Use when the user says: «it doesn''t work», «still broken», «hung/stuck», «disappeared after the update», «it used to work», «the metric is zero but the UI is fine» — or when an incident needs to be analyzed, the root cause of a hang found, and whether a process was actually restarted verified. Covers: facts before theories (storage/logs/PID), symptom vs root cause, silent failure, restart ritual, localizing hangs with a progress marker, timeouts, cache masking. Do not use for writing tests (testing-discipline).'
license: MIT
compatibility: 'any stack: processes, logs, DB, network, tests'
metadata:
  version: "4.7.0"
---

# Debug & incident protocol: facts before theories

## 1. Core: facts, not opinions

1. **FACTS BEFORE THEORIES** — order: 1) config flags, 2) DB row, 3) log lines, 4) only then a code hypothesis. The first incident report contains a fact, not an opinion.
2. **SYMPTOM ≠ ROOT CAUSE** — «minutes are not deducted» is a symptom; silent except + wrong parser is the root cause. Trace the call path to the side effect; fix the root cause ONCE.
3. **IF METRIC FLAT WHILE FEATURE «WORKS» — SILENT FAILURE** — UX success + zero metric = swallowed error. Look for except/early return on the metric path.
4. **SINGLE CONSUMER FOR EXCLUSIVE STREAMS** — long-poll/queue/lock file — one owner. Kill duplicates before start; health = exactly one PID.
5. **RESTART RITUAL IS PART OF THE FIX** — code on disk ≠ code in memory. After the fix: stop all → start one → verify log. Check: PID creation time > edit time.
6. **ENCODING OF CONSOLE ≠ ENCODING OF PRODUCT** — mojibake in the console does not mean corrupted data. Check UTF-8 in the client/file.

## 2. Localizing hangs

1. **«It can't hang» — not an argument.** Anything can hang: a network without a timeout, an interactive prompt, an infinite loop.
2. **Localize with a progress marker** — the marker («=== stage N ===») is printed at the START of the block → hang is INSIDE the last block with a marker.
3. **Progress — to a FILE, not a pipe** — `cmd 2>&1 | Out-File prog.txt`; a pipe can hang or lose the tail.
4. **Isolate the suspect BEFORE the full run** — one block with a small timeout (30-60s), not the whole set with 900s.
5. **Chain A && B && C masks the hang location** — split stages: each command separately, its own timeout.
6. **First suspect — infrastructure, not logic** — BeforeAll/modules/network/interactivity is guilty more often than an «instant» test.
7. **Compare with the last successful run** — git diff: the culprit is usually in the changes.

## 3. Cache and «still broken»

- **Old cache looks like an unfixed bug** — first check that the server IS SERVING the fixed bytes (curl + grep of the fix marker line), then force the cache to die.
- **Foreign processes on the port** — a test instance on the port blocks the real launch → looks «broken».

## Workflow (order of application)

1. **Gather facts, not theories.** config flags → DB row → log lines → code hypothesis.
2. **Rule out cache and foreign processes.** curl the served bytes + grep the fix marker. Foreign process on the port?
3. **Separate symptom from root cause.** Trace the call path to the side effect; fix the root cause ONCE.
4. **Check processes.** One consumer. PID trace chain: ParentProcessId + CreationDate + CommandLine. PID older than the edit = process on old code → restart ritual.
5. **Localize the hang with a marker.** Marker at the start of the block. Progress to a FILE. Isolate with a small timeout.
6. **Fix + verification.** stop-all-then-start-one → check fresh log lines → smoke the scenario.

## Incident checklist

- [ ] fact: config flags → DB row → log lines (before hypotheses)
- [ ] symptom separated from root cause; root cause fixed once
- [ ] progress markers present; stages separated
- [ ] timeout on every suspect
- [ ] one consumer per exclusive stream; PID new (CreationDate > edit)
- [ ] log checked by fresh lines after restart
- [ ] cache ruled out (served bytes verified)
- [ ] conclusion: fact + root cause + one fix + smoke

Attribution

oleg494oleg494
View sourceSee grades on GitHubMore from oleg494 →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Terse caveman voice: answer first, fluff gone, every technical fact kept. Use for /caveman, "caveman mode", "talk like caveman", "be brief", "less tokens". Stays on until "stop caveman" or "normal mode".

1100021 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

698461 votes

Writing Skills

Create and manage Claude Code skills in HASH repository following Anthropic best practices. Use when creating new skills, modifying skill-rules.json, understanding trigger patterns, working with hooks, debugging skill activation, or implementing progressive disclosure. Covers skill structure, YAML frontmatter, trigger types (keywords, intent patterns), UserPromptSubmit hook, and the 500-line rule. Includes validation and debugging with SKILL_DEBUG. Examples include rust-error-stack, cargo-dep...

3931 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3421 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Amp, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Grok Build, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

741 votes
View all in ai-agents →