From atlas
Reproducible bug, exception, stack trace, or bad output: root-cause fix with evidence, not a patch over. Use when you want the actual cause fixed, not the symptom hidden.
How this skill is triggered — by the user, by Claude, or both
Slash command
/atlas:atlas-debug [context] [stack] [symptom] [paste error/log]When to use
a reproducible bug, exception, stack trace, or bad output needs a root-cause fix with evidence, not a patch over
[context] [stack] [symptom] [paste error/log]This skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Apply the Operating Contract to this entire task. It is injected below.
Apply the Operating Contract to this entire task. It is injected below.
cat "${CLAUDE_PLUGIN_ROOT}/references/operating-contract.md"
If the contract did not load above, read ${CLAUDE_PLUGIN_ROOT}/references/operating-contract.md and apply it before proceeding.
Read ${CLAUDE_SKILL_DIR}/references/debug-workflow.md and follow the reproduce-localize-fix-verify loop it defines for every debug run.
Debug and fix this: $ARGUMENTS
Read the arguments as four inputs:
If the symptom or stack is missing detail you need, ask once for it, then proceed.
Steps:
atlas-loop skill to select and instantiate the best-fit loop from the loop-library, then run that loop. Otherwise, for non-trivial single-pass work, dispatch the squad rather than doing it all inline: dispatch all independent jobs in ONE message (multiple Agent calls in a single message) so they run concurrently, roughly 4-6 in flight - atlas:explorer to locate the failing path and its callers, debugger to confirm the root cause, atlas:implementer to apply the fix. ALWAYS close the wave with an independent atlas:verifier in a fresh context before integrating results.VERIFY:
REPORT:
npx claudepluginhub w159/atlas --plugin atlasGuides completion of development work by verifying tests, detecting environment, and presenting structured options for merge, PR, or cleanup.
Enforces test-driven development: write failing test first, then minimal code to pass. Use when implementing features or bugfixes.
Guides creation and editing of skills using test-driven development with pressure scenarios and subagents to verify agent compliance.