T2a
Skill t2a-protocol/T2A/integrations/t2a-memory/templates/skills/t2a
Tool-to-Agent Protocol: tools can be smart without embedded LLM calls.
npx -y skills add t2a-protocol/T2A --skill t2aAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
What its author says it does
Copied from the file, not written here
Complete T2A Protocol capsules when a tool returns t2a.handoff, points to task.json, needs output.json, or asks for validation.
SKILL.md
1.4 KB, as published. Nobody here has run it
T2A
Instructions
Use this skill when a tool returns type: "t2a.handoff", points to task.json, asks for output.json, or asks you to complete a T2A task.
- Read
task.jsonfirst. - Read
prompt.mdand declared input files. - Follow
output.schema.json. - Write structured output to the declared output path.
- Run validation if it is safe and available.
- Repair
output.jsonuntil validation passes. - Stop after validation unless the user explicitly asks you to resume or continue.
Guidelines
- Do not call external LLM APIs for T2A.
- Do not ask for
model,base_url,api_key, or provider configuration. - Do not invent IDs, URLs, citations, resource IDs, or references.
- Do not read undeclared files unless the user explicitly authorizes it.
- Missing
output.jsoncan be normal before you complete a pending handoff. - Treat T2A core validation and issuing-tool domain validation as separate layers.
- Do not resume side-effecting workflows unless explicitly instructed by the user, task policy, or host runtime.
Examples
- A CLI prints
T2A handoff requiredand returns a task directory. - A tool returns
{"type":"t2a.handoff","local":{"task_file":".../task.json"}}. - The user asks you to complete a T2A capsule and run validation.