Executes named subagent roles with optimized task instructions via explicit invocation.
npx skills add https://github.com/mystilleef/spae-framework --skill run-mystilleef
Please help me install this Agent Skill. Skill: run Source: https://github.com/mystilleef/spae-framework/tree/main/codex/skills/run Command: npx skills add https://github.com/mystilleef/spae-framework --skill run-mystilleef
$run, never triggering implicitly, which prevents unintended agent execution. - Task Optimization: Strips conversational fluff from instructions and rewrites them in precise, token-efficient language before dispatch. - Subagent Orchestration: Parses the first argument as the subagent name, spawns it with the optimized task, waits for completion, and presents results verbatim. - Use Case: In the SPAE workflow, invoke /run spec <proposal> to dispatch the spec agent, then /run plan, /run build, and other phase agents in sequence, each receiving clean, focused instructions. ## Quick Start Invoke the run skill with an agent name and task, such as asking it to run the spec agent on your feature proposal.High-intent search queries and answers about installing and using this skill.
Invoke it explicitly with $run followed by the agent name and task, for example $run spec my-feature-proposal. The first word after the invocation names the subagent, and everything after becomes the task instructions.
It strips conversational fluff and rewrites the task in precise, direct language for token and context efficiency before spawning the subagent. If no task is provided, it proceeds with an empty prompt.
No. The skill is configured with allow_implicit_invocation set to false in its policy, so it only executes when explicitly called with $run and never triggers implicitly.
The orchestrator waits for the subagent to finish, presents the result directly without preamble or commentary, and then stops. It performs no further operations and never modifies results unless explicitly requested.
In the SPAE framework it dispatches phase agents including spec, plan, inspect, build, check, fix, and verify. Each agent reads its designated inputs and writes its designated outputs within the seven-step workflow.
praxstackExpand SPEC.md into a detailed BLUEPRINT.md with validation rules and edge cases.
ATTCKDigitalConvert plan.md and spec.md into a dependency-ordered tasks.md document.
natslashRun a structured 5-phase plan review with triage and decision logging.
BeastAyyGAnalyze user requests and route them to suitable specialist agents.