一键在 Manus 中运行任何 Skill

$pwd:

oneshotprompt-cli

Name: Oneshotprompt Cli
Author: jihadkhawaja

// Operate the OneShotPrompt CLI predictably for validation, job discovery, targeted execution, and publishing. Use when a task is specifically about this project's command surface, config-driven runs, or the published console binary.

在 Manus 中运行

$ git log --oneline --stat

stars:0

forks:0

updated:2026年4月5日 18:10

SKILL.md

readonly

name	oneshotprompt-cli
description	Operate the OneShotPrompt CLI predictably for validation, job discovery, targeted execution, and publishing. Use when a task is specifically about this project's command surface, config-driven runs, or the published console binary.
metadata	{"author":"oneshotprompt","version":"1.0"}

Use this skill when the task is about interacting with OneShotPrompt itself rather than an arbitrary external CLI.

Workflow:

Identify whether the goal is help, config validation, job listing, running all jobs, running one job, keeping one job listening for WhatsApp bridge events, opening the interactive menu, or publishing the binary.
Prefer RunDotNetCommand when operating from a source checkout. Use RunCommand only when the task is explicitly about a published executable.
Pass the subcommand and arguments explicitly. Do not rely on shell syntax, redirection, or implicit defaults that depend on terminal interactivity.
In automation or scripted contexts, pass --config <path> explicitly instead of relying on the default config.yaml.
Use validate --config <path> before run when the config may be new or recently edited.
Use jobs --config <path> before run --job <name> when the exact job name is not already confirmed.
Use run --config <path> --job <name> for a targeted run and run --config <path> for all enabled jobs.
Use listen --config <path> --job <name> when the task is specifically about waiting for WhatsApp personal channel events and re-running one selected job each time a new allowlisted message arrives.
Use interactive or -i only when the task explicitly calls for menu-driven or ad-hoc prompting behavior.
Summarize the actual command outcome, including any validation errors, selected job, listener state, logs path, or publish output path.

CLI contract:

Supported commands are run, listen, validate, jobs, interactive, and help.
--job is valid with run and required with listen.
With no arguments, the app defaults to run --config config.yaml, but interactive terminals may open the menu instead. Agents should prefer explicit subcommands.
logs/, sibling skills/, and .oneshotprompt/memory/ are all resolved relative to the active config file location.
Native AOT publish output is written under src/OneShotPrompt.Console/bin/Release/net10.0/<rid>/publish/.

Guardrails:

Do not invent unsupported flags or subcommands.
Do not claim a job ran, a config validated, or a publish succeeded unless the command output confirms it.
Prefer the dedicated filesystem skills for simple reads or writes around the CLI instead of wrapping everything in command execution.
If the task depends on interactive menu choices, say that human input is required instead of pretending the menu was completed automatically.

related-skills.json

同仓库

whatsapp-personal-channel.md

from "jihadkhawaja/OneShotPrompt"

Operate the local unofficial WhatsApp personal channel bridge for a personal account, limited to the allowlisted phone number or numbers in the bridge config.

2026-04-050

oneshotprompt-cli.md

from "jihadkhawaja/OneShotPrompt"

Repository guidance for agents that need to communicate with the OneShotPrompt CLI, validate configs, list jobs, run specific jobs, or publish the console app.

2026-04-050

file-system-mutations.md

from "jihadkhawaja/OneShotPrompt"

Plan and perform minimal local filesystem changes after inspection. Use when a task requires creating folders, moving or copying files, deleting files, or writing text files.

2026-03-080

professional-software-engineering.md

from "jihadkhawaja/OneShotPrompt"

Internal project-agent engineering workflow for implementing features and fixes with unit tests, clean abstractions, extensible design, repeatable setup, cross-platform execution, and Docker-ready delivery. Use when the agent is building, refactoring, hardening, testing, packaging, or operationalizing software professionally.

2026-03-070

command-execution.md

from "jihadkhawaja/OneShotPrompt"

Run installed executables directly for cross-platform automation. Use when a task needs CLI-based work, scripted workflows, repository commands, or tool-driven operations that are broader than simple file reads and writes.

2026-03-070

file-system-inspector.md

from "jihadkhawaja/OneShotPrompt"

Inspect and understand local files and directories before planning or making changes. Use when a task mentions files, folders, paths, Downloads, Documents, Desktop, reading text files, or checking current filesystem state.

2026-03-070

package.json

"author": "jihadkhawaja"

"repository": "jihadkhawaja/OneShotPrompt"

打开 GitHub 仓库查看创作者相关仓库

$ install --global

$ download --local

在 Manus 中运行

$ useful --forSOC

软件开发工程师计算机与数学类职业15-1252L4

name	oneshotprompt-cli
description	Operate the OneShotPrompt CLI predictably for validation, job discovery, targeted execution, and publishing. Use when a task is specifically about this project's command surface, config-driven runs, or the published console binary.
metadata	{"author":"oneshotprompt","version":"1.0"}

Use this skill when the task is about interacting with OneShotPrompt itself rather than an arbitrary external CLI.

Workflow:

Identify whether the goal is help, config validation, job listing, running all jobs, running one job, keeping one job listening for WhatsApp bridge events, opening the interactive menu, or publishing the binary.
Prefer RunDotNetCommand when operating from a source checkout. Use RunCommand only when the task is explicitly about a published executable.
Pass the subcommand and arguments explicitly. Do not rely on shell syntax, redirection, or implicit defaults that depend on terminal interactivity.
In automation or scripted contexts, pass --config <path> explicitly instead of relying on the default config.yaml.
Use validate --config <path> before run when the config may be new or recently edited.
Use jobs --config <path> before run --job <name> when the exact job name is not already confirmed.
Use run --config <path> --job <name> for a targeted run and run --config <path> for all enabled jobs.
Use listen --config <path> --job <name> when the task is specifically about waiting for WhatsApp personal channel events and re-running one selected job each time a new allowlisted message arrives.
Use interactive or -i only when the task explicitly calls for menu-driven or ad-hoc prompting behavior.
Summarize the actual command outcome, including any validation errors, selected job, listener state, logs path, or publish output path.

CLI contract:

Supported commands are run, listen, validate, jobs, interactive, and help.
--job is valid with run and required with listen.
With no arguments, the app defaults to run --config config.yaml, but interactive terminals may open the menu instead. Agents should prefer explicit subcommands.
logs/, sibling skills/, and .oneshotprompt/memory/ are all resolved relative to the active config file location.
Native AOT publish output is written under src/OneShotPrompt.Console/bin/Release/net10.0/<rid>/publish/.

Guardrails:

Do not invent unsupported flags or subcommands.
Do not claim a job ran, a config validated, or a publish succeeded unless the command output confirms it.
Prefer the dedicated filesystem skills for simple reads or writes around the CLI instead of wrapping everything in command execution.
If the task depends on interactive menu choices, say that human input is required instead of pretending the menu was completed automatically.

oneshotprompt-cli

同仓库更多 Skills

同仓库更多 Skills