Menu
Use when modifying sparks plugin, debugging hooks, adding knowledge categories, or understanding how knowledge capture works
Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.
Baseado na classificação ocupacional SOC
Use this skill when you see @name or /name patterns in user input without file identifiers. This means the user is trying to call a subagent or slash command using Spectre Agent Tools. The @ prefix dispatches to a subagent (e.g., '@tdd-agent write tests', '@code-reviewer check this file'). The / prefix executes a slash command (e.g., '/spectre:scope', '/spectre:plan'). Both require Spectre CLI to run specialized agents and predefined prompts.
Use when modifying build loop code, debugging stats/token tracking, adding CLI features, changing iteration prompts, or understanding how spectre-build works end-to-end
Use when modifying the plan loop, debugging plan stages, changing clarification flow, scope isolation, or understanding how spectre-build --plan works end-to-end
Use when modifying the ship loop, debugging ship stages, changing clean/test/rebase behavior, or understanding how spectre-build --ship works end-to-end
Use when user wants to search for existing knowledge, recall a specific learning, or discover what knowledge is available.
Use when planning product scaling, adding GUI/server layers, multi-model support, adversarial reviews, live steering, telemetry, scheduling/triggers, or node-based pipeline editors to spectre-build
| name | feature-sparks-plugin |
| description | Use when modifying sparks plugin, debugging hooks, adding knowledge categories, or understanding how knowledge capture works |
| user-invocable | false |
Trigger: sparks, /learn, /find, knowledge, registry Confidence: high Created: 2025-01-18 Updated: 2025-01-18 Version: 2
Sparks is a knowledge capture and retrieval system for Claude Code projects. It solves the problem of context loss between sessions - when you debug something hard, discover a gotcha, or establish a pattern, that knowledge usually dies with the conversation.
Sparks captures durable knowledge as project-scoped skills that Claude automatically discovers and loads when relevant. Think of it as building institutional memory for your codebase.
| Problem | How Sparks Helps |
|---|---|
| Debugging the same issue twice | Gotchas captured, auto-loaded next time |
| "Why did we do it this way?" | Decisions + rationale preserved |
| New feature, no context | Feature dossiers explain architecture |
| Onboarding (human or AI) | Patterns and procedures documented |
| Knowledge siloed in conversations | Extracted into searchable, loadable skills |
After spending time debugging a tricky issue:
/learn the JWT refresh token was failing because...
Creates a gotcha that loads automatically when someone hits similar symptoms.
After implementing or understanding a feature:
/learn how the auth system works
Creates a feature dossier with flows, key files, and common tasks.
After making a significant choice:
/learn we chose Redis over Postgres for sessions because...
Preserves the rationale so future you (or teammates) don't reverse it blindly.
Before diving into unfamiliar code:
/find auth
Searches registry, loads relevant skills into context before you start.
/learn)/learn with optional topic/find or sparks-find skill)/find {query} or Claude loads sparks-find skillSkill({skill-name}){{project}}/.claude/skills/
├── sparks-find/
│ ├── SKILL.md # Find skill (generated, embedded registry)
│ └── references/
│ └── registry.toon # Registry source of truth
├── {category}-{slug}/ # Each learning = one skill
│ └── SKILL.md
└── ...
plugins/sparks/
├── skills/
│ ├── learn/
│ │ ├── SKILL.md # Capture workflow
│ │ └── references/
│ │ └── find-template.md # Template for project find skill
│ └── apply/
│ └── SKILL.md # Compliance (fallback for non-hook agents)
└── hooks/scripts/
├── load-knowledge.py # SessionStart - injects registry
└── register_spark.py # Creates registry + find skill
# registry.toon
{skill-name}|{category}|{triggers}|{description}
Example:
feature-auth-flows|feature|auth, login, JWT|Use when implementing or debugging authentication
gotchas-hook-timeout|gotchas|hook, timeout, silent|Use when hooks fail without error output
| Category | When to Use |
|---|---|
| feature | End-to-end feature knowledge (flows, architecture, key files) |
| gotchas | Hard-won debugging insights, non-obvious pitfalls |
| patterns | Repeatable solutions used across codebase |
| decisions | Architectural choices + rationale |
| procedures | Multi-step processes (deploy, release, migrate) |
| integration | Third-party APIs, vendor quirks |
| performance | Optimization learnings, benchmarks |
| testing | Test strategies, coverage decisions |
| ux | Design patterns, interaction conventions |
| strategy | Roadmap decisions, prioritization rationale |
user-invocable: false - skills don't clutter /skill list but agent loads via Skill toolregister_spark.py keeps find skill in sync| File | Purpose |
|---|---|
plugins/sparks/skills/learn/SKILL.md | Capture workflow - categorize, propose, write, register |
plugins/sparks/skills/learn/references/find-template.md | Template for generating project find skill |
plugins/sparks/skills/apply/SKILL.md | Compliance instructions for non-hook agents |
plugins/sparks/hooks/scripts/load-knowledge.py | SessionStart hook - reads registry, injects context |
plugins/sparks/hooks/scripts/register_spark.py | Creates/updates registry + regenerates find skill |
plugins/sparks/skills/learn/SKILL.md.claude/skills/sparks-find/references/registry.tooncd /project && python3 plugins/sparks/hooks/scripts/load-knowledge.pysystemMessage and additionalContext/learn in project with a topic.claude/skills/{category}-{slug}/SKILL.md.claude/skills/sparks-find/references/registry.toon