Mascot Daemon Skill

Version: 1.0.0  |  License: Apache-2.0

An Agent Skill that teaches AI coding assistants how to drive Mascot Daemon programmatically through its task database (TaskDB) — creating, submitting, and monitoring proteomics search tasks without the GUI.

The skill is just markdown — skills/mascot-daemon/SKILL.md plus reference files. It contains no Matrix Science software and redistributes nothing.

Mascot Daemon is a Windows application, so this skill targets a Windows environment.

What it covers

• The task lifecycle end to end: create a parameter set, insert a task, attach data files, hand it to the Daemon engine, poll for completion, read results back
• The full TaskDB schema — all five Mascot_Daemon_* tables, every parameter-set column, and the positional filter_options string the Distiller import filter is configured with
• The failure modes that actually bite: tasks that complete silently having done nothing, the next_action placeholder timestamp, task_status vs action_request ownership, the MAX(id)+1 race, UNC path requirements, instrument names that must match the server
• Recipes — re-searching an existing MGF to skip Distiller reprocessing, parallelising Distiller passes across tasks, reading Percolator-rescored results, and checking a finished search really is at the FDR you assumed

Daemon's engine keeps polling the same tables, so a task inserted by a script is picked up and run exactly as though a user had built it in the GUI. The skill drives Daemon rather than bypassing it.

Prerequisites

Beyond a working Mascot Daemon and Mascot Server, two configuration changes are required before anything in this skill will work. Both are one-off administrator jobs, and skills/mascot-daemon/references/SETUP.md walks through both in full.

1. The task database must be moved to a client/server SQL engine.

Mascot Daemon 2.5 defaults to VistaDB, an embedded single-file database with no network listener — nothing outside the Daemon process can connect to it. Daemon 2.4 and earlier defaulted to Microsoft Access, which is no better. Move TaskDB to PostgreSQL, Microsoft SQL Server, MySQL, or Oracle (Preferences → ODBC connection tab). Daemon's built-in XML backup/restore migrates existing task history between engines without loss.

Sites already sharing one task database across several Daemon instances will have done this.

2. Mascot Server security must be enabled.

The mascot_user_id column and every <mascot_user_*> template tag are only populated when the Mascot Server has security switched on. Daemon must also authenticate as a user in the power users or daemons group to be permitted to submit searches, with credentials entered in Preferences → Authentication → Mascot Security.

More on Mascot Daemon: https://www.matrixscience.com/daemon.html

Install — Claude Code

Pick one location depending on whether you want the skill globally or per-project.

Globally (all projects):

git clone https://github.com/Matrix-Science/mascot-daemon-skill.git
New-Item -ItemType Directory -Force "$env:USERPROFILE\.claude\skills" | Out-Null
Copy-Item -Recurse mascot-daemon-skill\skills\mascot-daemon "$env:USERPROFILE\.claude\skills\"

Per project (run from your project's root):

New-Item -ItemType Directory -Force .claude\skills | Out-Null
Copy-Item -Recurse "C:\path\to\mascot-daemon-skill\skills\mascot-daemon" .claude\skills\

If you'd rather not use the command line, clone or download the repo in File Explorer and copy the skills\mascot-daemon folder into either:

• %USERPROFILE%\.claude\skills\ (global), or
• <your-project>\.claude\skills\ (per-project)

After install

Restart Claude Code (or start a new session). The skill auto-loads when you mention Daemon tasks, batch searching, task submission, peak list processing, or the Mascot Daemon database. You can also force it with /skill mascot-daemon.

Install — other AI coding tools

The skill format follows Anthropic's Agent Skills spec, which several tools now support:

Tool	Supported	Install location
Claude Code (CLI / desktop / IDE plugins)	yes	%USERPROFILE%\.claude\skills\ or .claude\skills\ (project)
Claude API / Agent SDK	yes	Pass via the Skills API or bundle with your agent

For tools that don't yet support the skill format, the SKILL.md is just a long-form prompt:

1. Paste it in — open skills/mascot-daemon/SKILL.md and paste the contents into the chat or system prompt.
2. Reference it as context — most assistants let you @-mention or attach files. Point at skills/mascot-daemon/SKILL.md and the relevant references/*.md.

Using the skill

Once installed, just ask in plain language. Examples:

• "Submit these 40 raw files to Daemon as a batch using the PXD028735 parameter set."
• "Task 214 is paused — find out why."
• "Re-run that search with Percolator on, but reuse the MGF instead of re-running Distiller."
• "Show me every task from the last week that finished with no results."
• "Set up a parameter set for tryptic SwissProt searches at 10 ppm with a decoy database."

Repo layout

skills/mascot-daemon/
  SKILL.md                          # main skill entry point — prerequisites, connection,
                                    #   task lifecycle, submission code, gotchas
  references/
    SETUP.md                        # the two prerequisites in full: choosing and creating a
                                    #   SQL task database, ODBC drivers and connection settings,
                                    #   migrating existing history, Mascot Server security
    SCHEMA.md                       # all five TaskDB tables column by column, the positional
                                    #   filter_options string, template tags, legacy .par format
    RECIPES.md                      # query patterns, MGF reuse, parallel Distiller passes,
                                    #   Percolator results, FDR sanity check

Credentials

The skill reads database and Mascot credentials from environment variables and never hardcodes them. Keep them in a gitignored .env file or the platform credential store. The database login used by scripts needs only SELECT, INSERT and UPDATE on the mascot_daemon_* tables — not the create/drop rights Daemon itself needed when it first built them.

Related

• Matrix-Science/mascot-parser-skill — reads Mascot .dat results files with msparser. The natural next step once a Daemon task has produced results.
• Matrix-Science/mascot-mdro-skill — drives the Mascot Distiller SDK for raw data access and peak list generation, covering what Daemon's Distiller import filter does for you here.
• Matrix-Science/mascot-distiller-skill — builds custom Mascot Distiller quantitative reports.

License

The skill content in this repository (the markdown under skills/ and the documentation at the repo root) is licensed under the Apache License, Version 2.0 (Apache-2.0). See LICENSE for the full text.

Mascot Daemon, Mascot Server and Mascot Distiller are not covered by this license — they are proprietary software licensed separately by Matrix Science. See https://www.matrixscience.com/daemon.html for licensing.