portunus-prompt
Skill2 starsChanged 32 days agoMITView on GitHub
Have the HUMAN type a secret (sudo password, master credential, anything they haven't pasted to you) straight into Portunus through a desktop dialog or terminal window, without the value ever entering your context. Use whenever someone wants to store a secret "through the agent" that they have not already shared in chat.
In this repository6 other files it gives its agents.
- ui/AGENTS.mdAGENTS.md
- ui/CLAUDE.mdCLAUDE.md
- portunus-askSkill
- portunus-dropSkill
- portunus-vault-auditSkill
- portunus-vault-setupSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 63
Never the value, its length or a hash, on any path.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Portunus Prompt
Thin wrapper around portunus_prompt (MCP) or portunus prompt (CLI). You trigger it;
the human answers in a window on their own desktop; Portunus stores the answer in the
local vault. All you get back is metadata. Use this instead of portunus-drop whenever
you do not already hold the value.
When to use
- "Let me store my sudo password through you", "save my master password in Portunus", "I'll type the key in, don't look at it".
- Any credential the human hasn't pasted into the conversation. Don't ask them to paste
it so you can use
portunus_drop, and don't ask them to type a shell one-liner.
When NOT to use
- The human already pasted the value to you: use
portunus-drop(the value is already in your context, prompting again adds nothing). - There's no desktop session (headless server, SSH without X forwarding): the tool
returns an error saying so. Tell the human to run
portunus drop <name> <sm_name>at their own terminal, which prompts masked and twice.
Usage
portunus_prompt(
name="sudo-pw", # reference name
sm_name="dostal-sudo", # vault key
provider="linux", # optional metadata
description="sudo on workstation", # shown to the human in the prompt
enable=True, # optional: make it injectable right away
gate=True, # optional: require `portunus approve` per use
)CLI: portunus prompt sudo-pw dostal-sudo --provider linux --description "sudo on workstation" --enable --gate
(prints one JSON object; exit 0 stored, 2 cancelled, 1 error).
The call blocks until the human answers, cancels or the timeout (default 300 s) passes. Tell the human a window is about to open before calling it.
kind defaults to password, which keeps leading/trailing whitespace exactly as typed.
confirm=True asks twice in dialog surfaces; terminal windows always ask twice.
Surfaces (auto-selected in this order): macOS osascript dialog; Linux kdialog,
zenity, a GUI pinentry; then a new terminal window (konsole, gnome-terminal, xterm,
Terminal.app) running a masked prompt. surface="ui" instead creates a requested
placeholder and opens the dashboard's fulfill page; it returns {"pending": true}
immediately, so check portunus status <name> later.
What you'll see
{"name", "state", "stored": true, "gated", "surface"}: stored. Report it by name.{"cancelled": true}(maybe with"reason": "timeout"): nothing stored. Ask the human whether to try again; don't retry in a loop.{"error": "..."}: nothing stored; the message says why.
Never the value, its length or a hash, on any path.
Using it afterwards (e.g. sudo)
Pair a gated reference with a wall-clock approval: the human runs
portunus approve sudo-pw --ttl 2h, and the reference resolves for two hours
(a bare number such as --ttl 3 still means audit-clock ticks).
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

