breadcrumb
Leave notes on files for other agents to see in future sessions. Use after making non-obvious changes, fixing tricky bugs, or when code looks wrong but is intentional.
Best use case
breadcrumb is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Leave notes on files for other agents to see in future sessions. Use after making non-obvious changes, fixing tricky bugs, or when code looks wrong but is intentional.
Teams using breadcrumb should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/breadcrumb/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How breadcrumb Compares
| Feature / Agent | breadcrumb | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
Leave notes on files for other agents to see in future sessions. Use after making non-obvious changes, fixing tricky bugs, or when code looks wrong but is intentional.
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
SKILL.md Source
# Breadcrumb
Leave notes on files that persist across agent sessions.
## When to use
After making changes that future agents might misunderstand:
- Non-obvious code that looks like it could be simplified
- Bug fixes for edge cases
- Intentional workarounds
- Security-critical patterns
- Performance tuning
## Core workflow
**1. Before editing, check for warnings:**
```bash
breadcrumb check ./src/api/users.ts
```
- Exit 0 = safe to proceed (clear/info)
- Exit 1 = warning exists, read the `suggestion` field
**2. After non-obvious changes, leave a note:**
```bash
breadcrumb add ./src/api/users.ts "Retry logic tuned for rate limits"
```
## Command reference
| Command | Purpose |
|---------|---------|
| `breadcrumb check <path>` | Check path for notes (`-r` for recursive) |
| `breadcrumb add <path> <message>` | Leave a note (`-s` severity, `--ttl` expiration) |
| `breadcrumb edit <path-or-id>` | Edit a note (`-m` message, `-a` append, `-s` severity) |
| `breadcrumb verify [path]` | Check if notes are stale (`--update` to refresh hashes) |
| `breadcrumb search <query>` | Find notes by content (`-r` for regex) |
| `breadcrumb coverage [path]` | Show breadcrumb coverage stats |
| `breadcrumb ls` | List all notes (`-s` filter by severity) |
| `breadcrumb status` | Quick overview (counts) |
| `breadcrumb rm <path>` | Remove a note (`-i` by ID) |
| `breadcrumb prune` | Remove expired notes |
## Staleness detection
Notes track file content hashes. When you see `[STALE]` prefix:
- The file has changed since the note was written
- The note may no longer be accurate
- Use judgment: the warning might still apply, or might be outdated
```
📝 BREADCRUMB: [STALE] Don't simplify this regex
↑ Code changed - verify note still applies
```
After reviewing stale notes, update hashes with:
```bash
breadcrumb verify --update
```
## Output format
All commands output JSON. Key fields:
- `status`: "clear", "info", or "warn"
- `suggestion`: Actionable guidance when warnings exist
- `breadcrumbs`: Array of matching breadcrumb objects
- `staleness`: "verified", "stale", or "unknown" (per breadcrumb)Related Skills
bgo
Automates the complete Blender build-go workflow, from building and packaging your extension/add-on to removing old versions, installing, enabling, and launching Blender for quick testing and iteration.
self-subagent
Orchestrate parallel sub-tasks by spawning non-interactive instances of your own CLI as subagents. Use when you need to parallelize work across multiple files, run independent investigations simultaneously, or delegate heavy multi-step tasks. Works with ANY AI coding CLI agent (Amp, Claude Code, Codex, Cursor, OpenCode, aider, Cline, Roo, goose, Windsurf, Copilot CLI, pi, etc.). Triggers on "run in parallel", "subagent", "delegate", "fan out", "concurrent tasks", or any complex task that benefits from parallel execution.
self-improving-ai
Understanding and using StickerNest's self-improving AI system. Use when the user asks about AI self-improvement, prompt versioning, reflection loops, AI evaluation, auto-tuning prompts, or the AI judge system. Covers AIReflectionService, stores, and the improvement loop.
second-brain
Personal intelligence system for capturing thoughts, managing knowledge, and surfacing insights. Use when user wants to capture an idea, task, or note during conversation; query their knowledge base; check their inbox; review digests; or update task status. Triggers include "remember this," "add a task," "what did I say about," "show my inbox," or "mark complete."
searching-message-history
Search Telegram conversation history and stored links. Use when finding past messages, what someone said, or links shared in chats.
screpcombiningexpression
Combine scTCR/BCR repertoire data with scRNA-seq expression data using `scRepertoire::combineExpression()`. This process integrates immune receptor information (CDR3 sequences, V(D)J genes, clonotypes) into a Seurat object's metadata, enabling clonotype-aware gene expression analysis.
scrapegraph-ai-automation
Automate Scrapegraph AI tasks via Rube MCP (Composio). Always search tools first for current schemas.
scientific-schematics
Create publication-quality scientific diagrams using Nano Banana Pro AI with smart iterative refinement. Uses Gemini 3 Pro for quality review. Only regenerates if quality is below threshold for your document type. Specialized in neural network architectures, system diagrams, flowcharts, biological pathways, and complex scientific visualizations.
scientific-papers-to-dataset
Build structured datasets from academic papers. Use when the user wants to extract structured data from scientific literature, traverse citation graphs, search OpenAlex for papers, or create datasets from PDFs for research purposes.
scholarag
Build PRISMA 2020-compliant systematic literature review systems with RAG-powered analysis in VS Code. Use when researcher needs automated paper retrieval (Semantic Scholar, OpenAlex, arXiv), AI-assisted PRISMA screening (50% or 90% threshold), vector database creation (ChromaDB), or research conversation interface. Supports knowledge_repository (comprehensive, 15K+ papers, teaching/exploration) and systematic_review (publication-quality, 50-300 papers, meta-analysis) modes. Conversation-first workflow with 7 stages.
savestate
Time Machine for AI. Encrypted backup, restore, and cross-platform migration for your agent's memory and identity. Supports OpenClaw, ChatGPT, Claude, Gemini, and more. AES-256-GCM encryption with user-controlled keys.
sarvam-ai-skills
Guide for building AI applications with Sarvam AI APIs for Indian languages. Use when working with speech-to-text transcription, text-to-speech synthesis, text translation, chat completion, or document intelligence. Covers models saarika:v2.5, saaras:v2.5/v3, bulbul:v3, mayura:v1, sarvam-translate:v1, sarvam-m, and sarvam-vision for 11-23 Indian languages. Trigger when user asks about Indian language AI, STT, TTS, translation, multilingual chatbots, voice assistants, or document processing.