Loads a user's mined Ditto profile so Openclaw Skills agents follow their real working style instead of starting from zero.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install ditto-profile
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install ditto-profile using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Ditto Profile Loader?
Ditto Profile is an Openclaw Skills workflow that resolves a user's private working profile from local Claude Code, Codex, and OpenCode session logs. Instead of relying on a generic agent persona, it loads the user's own guidance for what done looks like, what gets rejected immediately, how they debug, and how they write.
Use it at the start of a task to avoid a cold start and keep the agent aligned with the user's actual preferences. The profile stays on the user's machine; Openclaw Skills only asks ditto-cli for the active profile paths, then reads those files directly and fails fast if no active profile exists.
Ditto Profile Loader Use Cases
- Starting any coding task where the user wants the assistant to behave like their usual self.
- Before implementation, debugging, refactoring, or review when task quality depends on their personal definition of done.
- When the user prefers specific writing, reasoning, or debugging habits that should override default agent behavior.
- When you need to detect and stop on missing, stale, or corrupt profile pointers instead of guessing.
- When you want Openclaw Skills to bootstrap with user-specific context from local session logs, not a generic prompt.
- When the user has not yet mined a profile and needs to run
uvx ditto-clifirst.
How Ditto Profile Loader Works
- Run
uvx ditto-cli plugin profile-path --domain workto resolve the active Ditto profile path JSON. - If the command exits nonzero with
no active Ditto profile, stop and instruct the user to runuvx ditto-clito mine their local logs into a private profile. - Read every returned file path completely; Openclaw Skills treats the contents as the user's actual working instructions for this task.
- Apply the profile as execution policy: honor their definition of done, rejection criteria, debugging approach, and writing style.
- Never continue from a stale, missing, or corrupt pointer; only proceed when the profile is current and readable.
Ditto Profile Loader Setup
- Make sure
uvis available souvxcan fetch and runditto-clifrom PyPI. - Resolve the work profile path.
uvx ditto-cli plugin profile-path --domain work
- If no active profile exists, initialize Ditto on the user's machine so it can mine local Claude Code, Codex, and OpenCode logs into a private profile.
uvx ditto-cli
- Re-run the profile-path command, then load every file path it returns before starting the task.
- No repository bundle or extra config is required in Openclaw Skills; the runtime is fetched on demand and the profile remains local.
Ditto Profile Loader Data Schema & Taxonomy
Ditto Profile does not create a project-local schema. It consumes the following local profile metadata and file references:
| Entity | Source | Shape | Purpose |
|---|---|---|---|
| Active profile pointer | ditto-cli plugin profile-path --domain work |
JSON output | Returns one or more file paths to the active profile |
| Profile file(s) | Local filesystem on the user's machine | One or more readable files | Hold the user's actual working instructions |
| Domain | CLI flag | String (work) |
Selects the work persona/profile |
| Status | CLI exit code and stderr | Success or no active Ditto profile |
Tells Openclaw Skills whether to proceed or stop |
| Behavioral instructions | Profile contents | Natural-language guidance | Defines done criteria, rejection rules, debugging style, and writing style |
| Privacy boundary | Local machine | Local-only data | Keeps mined data private; nothing is bundled into the skill |
Metadata taxonomy used by Openclaw Skills:
- Source systems: Claude Code, Codex, OpenCode
- Scope: work
- Integrity state: active, missing, stale, corrupt
- Instruction types: done criteria, reject rules, debugging habits, writing style
Generated files: none from this skill itself. The separate uvx ditto-cli bootstrap step may create the local private profile, but the skill only resolves and reads it.
Ditto Profile Loader Advanced Features
- Multi-source mining from local Claude Code, Codex, and OpenCode logs.
- Fail-fast validation when no active profile is available, so the agent never invents a persona.
- Local-only privacy model that keeps mined profile data on the user's machine.
- Execution aligned to the user's own done criteria, rejection patterns, debugging style, and writing style.
- Domain-aware resolution with
--domain workfor a clean separation of contexts. - Lightweight pre-task bootstrap that turns Openclaw Skills from a cold start into a personalized workflow.
SKILL.md
Loading
Related Openclaw Skills
MCP Precheck
Openclaw Skills for mcp-precheck gives you deterministic, registry-backed MCP server vetting before any connection, config edit, or hook install.
Journal Metadata Lookup
Journal Metadata Lookup is an Openclaw Skills utility that resolves a paper identifier into authoritative authorship, venue, and citation metadata in one command.
Morning Intelligence
An interview-driven Openclaw Skills workflow that turns your news preferences into a reusable morning briefing prompt.
Brag Doc Skill
Openclaw Skills turns scattered wins into a running, impact-first brag doc that is ready for reviews and promotion packets.
Tristan RFQ Overseer
Tristan RFQ Overseer is an Openclaw Skills RFQ pipeline that captures requests, prices them transparently, compares suppliers, and stages approved quotes for sending.
AKF — Agent Knowledge Format
AKF helps Openclaw Skills stamp, read, verify, and audit trust metadata in AI-generated files so provenance, confidence, and compliance travel with the content.
Featured*
