# Ctx

> Use when starting a session on a project, returning after time away, or before making significant changes. Essential for building comprehensive understanding of project state through total recall and deep exploration.

- Skill: `majiayu000/ctx` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds add majiayu000/ctx`
- Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/ctx/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: majiayu000 (https://skillmd.com/u/majiayu000)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/majiayu000/ctx

---


# Contextualize

You need to fully contextualize yourself in this repository before doing any work.

## Step 1: Search Total Recall

First, search Total Recall for relevant context about this project. Use whichever tool name is available:

```
# Claude Code:
mcp__plugin_totalrecall_memory-total-recall__synthesis_search({
  query: "project name/key terms recent work decisions made",
  max_results: 15
})

# Codex/other:
totalrecall.synthesis_search({
  query: "project name/key terms recent work decisions made",
  max_results: 15
})
```

Look for:
- Recent conversations about this project
- Decisions that were made
- Problems that were solved
- Patterns established
- Gotchas discovered

## Step 2: Dispatch the Oracle for Deep Context

**If Task tool is available**, use it to dispatch an Oracle:

```
Task(
  subagent_type: "general-purpose",
  model: "opus",
  prompt: """
  You are The Oracle - contextualize yourself deeply in this repository.
  [see investigation checklist below]
  """
)
```

**If Task tool is NOT available**, perform the investigation yourself directly.

### Investigation Checklist

Whether delegated to Oracle or done directly, cover these areas:

1. **Project Identity**
   - Read CLAUDE.md, README.md, package.json (or equivalent)
   - What is this project? What does it do?
   - What are the core technologies and patterns?
   - What are the established conventions?

2. **Current Work Status**
   - Check .plans/ directory for active plans
   - Check .tasks/ or TODO files for pending work
   - Check git status and recent commits (git log -10 --oneline)
   - What was being worked on? What's in progress?

3. **Decision History**
   - Check .plans/DECISIONS.md or .plans/decisions/
   - What major decisions have been made?
   - What constraints exist from past choices?

4. **Architecture Overview**
   - Map the key directories and their purposes
   - Identify core modules/services
   - Understand the data flow

5. **Pain Points & Context**
   - Look for TODOs, FIXMEs, HACKs in code
   - Check for any KNOWN_ISSUES or TROUBLESHOOTING docs
   - Identify technical debt markers

### Deliverable Format

```markdown
## Project Summary
[What this is, in 2-3 sentences]

## Current Status
- Active work: [what's in progress]
- Blocked items: [if any]
- Recent changes: [last few commits summary]

## Key Decisions to Remember
- [Decision 1]
- [Decision 2]

## Established Patterns
- [Pattern 1]: [where/how used]
- [Pattern 2]: [where/how used]

## Conventions & Constraints
- [Convention from CLAUDE.md]
- [Technical constraint]

## Recommended Next Steps
Based on current state, suggest what to focus on.
```

## Step 3: Synthesize and Report

After gathering context from memory and investigation:

1. **Summarize** the key findings for the user
2. **Highlight** anything that seems urgent or blocked
3. **Confirm** understanding of current priorities
4. **Ask** if there's specific context you're missing

## When to Use This Command

- Starting a new session on a project
- Returning to a project after time away
- Before making significant changes
- When feeling uncertain about project state
- After another developer has made changes

**Remember**: Context prevents mistakes. Take the time to understand before acting.