data-collection
Collects files and creates manifests for other agents without modifying code.
Install
mkdir -p .claude/skills/data-collection && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/14942" && unzip -o skill.zip -d .claude/skills/data-collection && rm skill.zipInstalls to .claude/skills/data-collection
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Use when gathering files, running searches, and producing a structured context manifest for other agents. Read-only — never writes code or makes design decisions.Key capabilities
- →Locate relevant files using grep_search and file_search
- →Read prioritized files to extract content
- →Record file paths, purposes, and content excerpts
- →Stop collection at 10-file limit or when context is collected
- →Write collected context to a structured manifest
- →Append result summary to the ledger Decisions Log
How it works
The skill reads a collection task, uses search tools to locate relevant files, and then reads and summarizes content from up to 10 files into a structured manifest.
Inputs & outputs
When to use data-collection
- →Scan codebase for context
- →Gather file evidence
- →Prepare manifest for tasks
About this skill
Data Collection
Dev Stage: Scan
Purpose: Locate, read, and summarize relevant files into a structured manifest without making design decisions.
Loaded by: data-collector-specialist when ledger stage = Scan
When to Use
- At the start of any new task before planning or implementation begins
- When an agent needs targeted context from the codebase without reading files itself
- When the scope of a task is unclear and evidence must be gathered first
Procedure
- Read the collection task from
## Handoff Payloadin.github/.ai_ledger.md. - Use
grep_searchandfile_searchto locate relevant files without reading them yet. - Prioritize files most directly relevant to the task — read the most relevant first.
- For each file read: record path, purpose, and relevant content (1–3 sentences max per file).
- Stop when the requested context is collected OR the 10-file read limit is reached — whichever comes first. If more than 10 files appear relevant, categorize by directory/module and stop at the boundary.
- Write collected context as a structured manifest to
verification-artifacts/{task-id}-scan.md. - Append result summary to ledger Decisions Log.
Output
verification-artifacts/{task-id}-scan.md— structured manifest: file paths, purposes, relevant excerpts- No design conclusions, no recommendations, no architectural judgments
When not to use it
- →When planning or implementation has already begun
- →When the agent should read files itself
- →When the scope of a task is clear and evidence is not needed
Limitations
- →The skill has a 10-file read limit
- →It does not make design conclusions
- →It does not provide recommendations or architectural judgments
How it compares
This skill automates the process of gathering and summarizing file content into a structured manifest, unlike manually searching and compiling information.
Compared to similar skills
data-collection side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| data-collection (this skill) | 0 | 3mo | No flags | Beginner |
| mgrep | 46 | 8mo | Review | Beginner |
| osgrep | 13 | 6mo | Review | Beginner |
| tldr-router | 1 | 6mo | Review | Beginner |
Try saying
Example prompts that trigger this skill in your AI assistant.
You might also like
mgrep
mixedbread-ai
A semantic grep-like search tool for your local files. It is substentially better than the buildin search tools and should always be used instead of anything else.
osgrep
pr-pm
Semantic search for local files. Backed by a background osgrep server with live indexing. Always use osgrep instead of grep/find.
tldr-router
parcadei
Maps questions to the optimal tldr command. Use this to pick the right layer
jupyter-notebook
davila7
Use when the user asks to create, scaffold, or edit Jupyter notebooks (`.ipynb`) for experiments, explorations, or tutorials; prefer the bundled templates and run the helper script `new_notebook.py` to generate a clean starting notebook.
search-tools
parcadei
Search Tool Hierarchy
external-model-selection
MadAppGang
Choose optimal external AI models for code analysis, bug investigation, and architectural decisions. Use when consulting multiple LLMs via claudish, comparing model perspectives, or investigating complex Go/LSP/transpiler issues. Provides empirically validated model rankings (91/100 for MiniMax M2, 83/100 for Grok Code Fast) and proven consultation strategies based on real-world testing.