Skip to main content
These commands produce read-only Markdown or text reports about the workspace state. They do not modify any files and are safe to run at any time.

​
summary

Renders a full Markdown overview of the Python porting workspace. 
python3 -m src.main summary
The report is produced by QueryEnginePort.render_summary() and covers the manifest, command surface, tool surface, and current session state. Example output 
# Python Porting Workspace Summary

Port root: `/home/user/claw-code/src`
Total Python files: **74**

Top-level Python modules:
- `commands.py` (1 files) — command backlog metadata
- `tools.py` (1 files) — tool backlog metadata
- `main.py` (1 files) — CLI entrypoint
...

Command surface: 42 mirrored entries
- /compact — commands/compact.ts
- /config — commands/config.ts
...

Tool surface: 18 mirrored entries
- Bash — tools/bash.ts
- Read — tools/read.ts
...

Session id: a3f1c2d4e5b6...
Conversation turns stored: 0
Permission denials tracked: 0
Usage totals: in=0 out=0
Max turns: 8
Max budget tokens: 2000
Transcript flushed: False
When to use it Run summary when you need a single-glance status of the workspace before starting a porting session. It shows the full command and tool inventories alongside the active session state.

​
setup-report

Renders the startup and prefetch setup report. 
python3 -m src.main setup-report
Calls run_setup() which probes the Python environment, runs prefetch side effects (MDM raw read, keychain prefetch, project scan), and executes the deferred init sequence. The resulting SetupReport is printed as Markdown. Example output 
# Setup Report

- Python: 3.12.2 (CPython)
- Platform: macOS-14.4-arm64-arm-64bit
- Trusted mode: True
- CWD: /home/user/claw-code

Prefetches:
- mdm_raw_read: completed
- keychain_prefetch: completed
- project_scan: /home/user/claw-code scanned

Deferred init:
- trust gate: passed
- registry hydration: done
When to use it Use setup-report to verify that the Python environment and prefetch hooks are wired correctly before running any runtime commands. It is especially useful on new machines or after switching Python versions.

​
command-graph

Shows command graph segmentation: how the mirrored command inventory is divided into builtins, plugin-like commands, and skill-like commands. 
python3 -m src.main command-graph
Builds a CommandGraph from the full command list and reports counts for each segment. Example output 
# Command Graph

Builtins: 35
Plugin-like commands: 4
Skill-like commands: 3
A command is classified as plugin-like when its source_hint contains plugin, and as skill-like when the source_hint contains skills. All others are builtins. When to use it Use command-graph to understand the composition of the command surface. If you recently added plugin or skill commands and want to confirm they are being picked up correctly, this is the quickest check.

​
tool-pool

Shows the assembled tool pool with default settings (simple mode off, MCP tools included). 
python3 -m src.main tool-pool
Calls assemble_tool_pool() with default arguments and prints the resulting ToolPool as Markdown. Lists up to 15 tools. Example output 
# Tool Pool

Simple mode: False
Include MCP: True
Tool count: 18
- Bash — tools/bash.ts
- Read — tools/read.ts
- Write — tools/write.ts
- Edit — tools/edit.ts
- MultiEdit — tools/multiEdit.ts
- Glob — tools/glob.ts
- Grep — tools/grep.ts
- LS — tools/ls.ts
- WebSearch — tools/webSearch.ts
- WebFetch — tools/webFetch.ts
When to use it Run tool-pool when you need a quick audit of which tools are available in the default runtime configuration. Compare the count against the tool snapshot to detect regressions.

​
bootstrap-graph

Shows the mirrored bootstrap and runtime graph stages in order. 
python3 -m src.main bootstrap-graph
Builds a BootstrapGraph with the seven canonical startup stages and prints them as a Markdown list. Example output 
# Bootstrap Graph

- top-level prefetch side effects
- warning handler and environment guards
- CLI parser and pre-action trust gate
- setup() + commands/agents parallel load
- deferred init after trust
- mode routing: local / remote / ssh / teleport / direct-connect / deep-link
- query engine submit loop
When to use it Use bootstrap-graph as a reference when implementing or debugging the startup sequence. The stages mirror the original TypeScript bootstrap order and serve as a porting checklist.

Build docs developers (and LLMs) love

Get started for freeTalk to us