Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →sapientinc/praxist-onboarding
Establish detailed context for Praxist before helping a user install, configure, operate, troubleshoot, or extend the system.
Passed 14 of 19 test cases with the skill, and 4 of 19 without it.
and 17 more in the full report
12 fixed5 passed either way6 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 12 test cases the skill fixed. Same model, same prompt, with and without the skill.
In current template-style task configurations, how frequently does the Directed Idea Generation (DIG) stage execute across research generations by default?
| Checked | Without skill | With skill |
|---|---|---|
| States that DIG runs for initial generation 0 only by default | Missed | Passed |
| Identifies the default scope configuration setting as dig_lite.generation_scope: initial_only | Missed | Passed |
The instructions your model reads when the skill loads.
Use this skill to give the current agent enough Praxist system context to be useful on a first interaction. Keep the explanation system-level: describe purpose, operating model, configuration surfaces, software boundaries, and key module locations. Do not paste large code blocks or task-specific benchmark logic into the conversation.
This skill is primarily for context establishment. If the user invokes only:
praxist-onboarding
or otherwise asks only to understand Praxist, do not start a run, initialize a task, edit files, change configuration, install dependencies, or perform any state-changing environment operation. Load the Praxist context, run the default read-only environment scan below, give a concise readiness note, and wait for the user's next instruction.
Use this shape for a bare invocation:
Praxist onboarding context loaded. I understand the system as a task-agnostic autonomous-research control plane with task projects, peers, generations, frontier/incubator/Gems, PI/Chair planning, plugin boundaries, and packaged docs/templates/skills. I will run a read-only environment scan, report the result, and then wait for your next instruction.
The read-only environment scan below is part of the default onboarding action. Only take write or runtime action when the user gives an explicit goal such as installing a skill, changing configuration, creating a task project, running praxist resolve, launching a run, inspecting a run deeply, or editing code.
When this skill is invoked, perform this read-only scan by default unless the user explicitly says not to scan. Do not install packages, modify shell config, start runs, write env files, edit task files, or change credentials during this scan.
Keep discovery bounded. If using ls, find, fd, rg --files, or similar filesystem discovery commands, limit the scan to explicitly selected roots and no more than two directory levels below each selected root. For default onboarding, the selected roots are the current directory and its immediate parent-level candidates only. Never recurse through /, $HOME, mount roots, shared-storage roots, full workspace roots, datasets, caches, or experiments/ trees during this scan.
Check these surfaces:
Praxist command and pip package
command -v praxistpraxist --help when praxist existspython -m pip show praxist when Python/pip are availablepython -c "import praxist; print(praxist.__file__)" to locate the imported packageSource checkouts near the current directory
pyproject.toml declares name = "praxist" and a praxist/ package directory exists.git status --short --branch.test cases passed, vs 4 / 19 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.