Slash Command

/explore

Starts interactive exploration of project transcripts: surveys sessions, tools, errors, interrupts; user steers targeted queries and saves findings.

Bash

Markdown

developer-tools

code-quality

npx claudepluginhub jamie-bitflight/claude_skills --plugin agentskill-kaizen

Details

Argument[--project <name>]

Allowed Tools

ReadWriteGlobGrepBashTask

Prompt Preview

Start an interactive transcript exploration session. Present initial findings from the transcript corpus, then let the user steer deeper investigation.

## Arguments

- `--project <name>` — Scope to a specific project transcript directory. Default: current project.

## Execution Steps

1. **Resolve transcript path.** Same logic as `/agentskill-kaizen:analyze` — derive project key from `--project` flag or current working directory.

2. **Run initial survey.** Use DuckDB MCP to run a quick corpus overview:
   - Total session count and date range
   - Record type distribution
   - Top 10 most-...

Command Content

Other plugins with /explore

/run

Runs full Self-Care triage pipeline on trace file (OTEL JSON or Claude Code JSONL), orchestrating validation, analysis, reporting, review, output lists, and optional feedback.

7 tools

self-care

/explore

57.5k

Guides Socratic ideation on an optional topic via probing questions, optional research, and routes outputs to GSD artifacts like notes, todos, seeds, requirements, or phases.

7 tools

gsd-build-get-shit-done

/explore

2.1k

Analyzes blockchain addresses, transactions, smart contracts, blocks, and tokens, producing structured markdown reports with balances, histories, holdings, and insights.

blockchain-explorer-cli

/explore

1.6k

Explores codebase to map structure, detect tech stack, patterns, entry points, and CI/CD setup, producing a structured project overview.

explore

/explore

692

Explores codebase iteratively by keyword: searches file names, code content, git history; refines results interactively or automatically up to depth N, tracks dependencies, summarizes structure and relations.

claude-forge

Stats

Parent Repo Stars43

Parent Repo Forks7

Last CommitFeb 19, 2026

Actions

View Source View Plugin View on GitHub View README

Help us improve

Share bugs, ideas, or general feedback.

Start an interactive transcript exploration session. Present initial findings from the transcript corpus, then let the user steer deeper investigation. ## Arguments - `--project <name>` — Scope to a specific project transcript directory. Default: current project. ## Execution Steps 1. **Resolve transcript path.** Same logic as `/agentskill-kaizen:analyze` — derive project key from `--project` flag or current working directory. 2. **Run initial survey.** Use DuckDB MCP to run a quick corpus overview: - Total session count and date range - Record type distribution - Top 10 most-...

Start an interactive transcript exploration session. Present initial findings from the transcript corpus, then let the user steer deeper investigation.

Arguments

--project <name> — Scope to a specific project transcript directory. Default: current project.

Execution Steps

Resolve transcript path. Same logic as /agentskill-kaizen:analyze — derive project key from --project flag or current working directory.
Run initial survey. Use DuckDB MCP to run a quick corpus overview:
- Total session count and date range
- Record type distribution
- Top 10 most-used tools
- Error rate summary
- User interrupt count
Present findings to user. Display the survey results and suggest investigation directions:
- "I found {N} sessions with {M} tool misuse violations. Want to dig into those?"
- "There are {K} user corrections. Want to see the frustration signals?"
- "The most common error is {type} ({count} times). Want to trace those sessions?"
Follow user direction. Based on user response, run targeted queries:
- Use DuckDB SQL for structured data extraction
- Use kaizen MCP tools for process mining and pattern detection
- Present results incrementally
- Ask clarifying questions to narrow investigation
Save findings on request. When the user wants to save findings, write to .planning/kaizen/exploration-{date}.md.

Interaction Pattern

This command runs interactively — do NOT spawn an autonomous agent. Stay in the main conversation so the user can steer the investigation in real-time. Use MCP tools directly for queries.