How this skill is triggered — by the user, by Claude, or both
Slash command
/ardent:compoundThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Capture a recently solved problem as a searchable solution document. Each documented solution compounds your knowledge — the first time you solve a problem takes research, the next time takes minutes.
Capture a recently solved problem as a searchable solution document. Each documented solution compounds your knowledge — the first time you solve a problem takes research, the next time takes minutes.
<context_hint> #$ARGUMENTS </context_hint>
Skip for: simple typos, obvious syntax errors, trivial fixes.
From the current conversation, identify:
If the conversation doesn't contain enough context, ask: "What was the problem you just solved? What was the root cause?"
First, determine scope: Is this solution specific to the current project, or is it general knowledge useful across any project?
Then determine the category:
| Category | When |
|---|---|
performance | Slow queries, memory issues, scaling |
debugging | Hard-to-find bugs, misleading errors |
configuration | Environment, tooling, build issues |
integration | External APIs, services, protocols |
architecture | Design decisions, patterns, boundaries |
testing | Test flakiness, mocking, test design |
electron | Electron-specific (IPC, security, packaging) |
typescript | Type system, compiler, patterns |
general | Anything else |
Derive {slug} from the problem context — a short slug describing the solution (e.g. n-plus-one-briefs, ipc-deadlock).
Derive {area} from the domain of work (see CLAUDE.md for area mapping) — used in frontmatter only, not in the path.
Output path depends on scope:
.untracked/solutions/YYYY-MM-DD-{slug}.md.untracked/.shared/{category}/YYYY-MM-DD-{slug}.mdCreate directories if they don't exist.
Project-specific template:
---
type: solution
project: {project}
area: {area}
topic: {slug}
date: YYYY-MM-DD
category: {category}
tags: [relevant, tech, keywords]
branch: {current branch if applicable}
status: complete
---
# {Descriptive title of the problem}
## Symptom
{What you observed — error messages, behavior, metrics}
## Root Cause
{Technical explanation of why this happened}
## Solution
{Step-by-step fix with code examples}
## What Didn't Work
{Investigation steps that were dead ends and why — saves future time}
## Prevention
{How to avoid this in the future — patterns, checks, tests}
Cross-project template:
---
type: resource
topic: {category}
date: YYYY-MM-DD
tags: [relevant, tech, keywords]
status: active
---
# {Descriptive title}
## Summary
{One-paragraph explanation of the gotcha/pattern/technique}
## Details
{Full explanation with code examples}
## References
{Links, docs, related solutions}
Output:
Solution documented at {path}
Key insight: {one-line summary of root cause and fix}
npx claudepluginhub ardent-ai/ardent-claude-plugins --plugin ardentCaptures verified solutions to non-trivial problems as categorized markdown files with YAML frontmatter in symptom directories for grep-searchable knowledge base. Auto-triggers on phrases like 'that worked' or /doc-fix command.
Captures verified solutions to bugs and issues as searchable Markdown docs in .claude/solutions/. Triggers after fixes, 'that worked', or /phx:review; searches duplicates first.
Coordinates subagents to document recently solved problems into searchable YAML-frontmatter Markdown files in docs/solutions/ while context is fresh.