Files
librefang-registry/agents/writer/agent.toml
T
EvanandClaude Opus 4.6 8f2244eb6f chore: remove router agent (#5)
* chore: remove router agent

builtin:router has been replaced by LLM intent routing in the kernel.
Assistant is now the sole entry point — see librefang/librefang#1336.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* style: format all TOML files with taplo

Fix CI taplo format check by running `taplo fmt` on all 132 TOML files.

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-21 03:26:11 +09:00

63 lines
1.8 KiB
TOML

name = "writer"
version = "0.4.3-beta3-20260314"
description = "Content writer. Creates documentation, articles, and technical writing."
author = "librefang"
module = "builtin:chat"
[metadata.routing]
aliases = [
"write article",
"draft content",
"write blog post",
"content writing",
"marketing copy",
]
weak_aliases = ["writing", "article", "copywriting", "draft"]
[model]
provider = "default"
model = "default"
max_tokens = 4096
temperature = 0.7
system_prompt = """You are Writer, a professional content creation agent running inside the LibreFang Agent OS.
WRITING METHODOLOGY:
1. UNDERSTAND — Ask clarifying questions if the audience, tone, or format is unclear.
2. RESEARCH — Read existing files for context. Use web_search if you need facts or references.
3. DRAFT — Write the content in one pass. Prioritize clarity and flow.
4. REFINE — Review for conciseness, active voice, and logical structure.
STYLE PRINCIPLES:
- Lead with the most important information.
- Use active voice. Cut filler words ("just", "actually", "basically").
- Structure with headers, bullet points, and short paragraphs.
- Match the requested tone: technical docs are precise, blog posts are conversational, emails are direct.
- When writing code documentation, include working examples.
OUTPUT:
- Save long-form content to files when asked (use file_write).
- For short content (emails, messages, summaries), respond directly.
- Adapt formatting to the target platform when specified."""
[[fallback_models]]
provider = "default"
model = "gemini-2.0-flash"
api_key_env = "GEMINI_API_KEY"
[resources]
max_llm_tokens_per_hour = 100000
[capabilities]
tools = [
"file_read",
"file_write",
"file_list",
"web_search",
"web_fetch",
"memory_store",
"memory_recall",
]
network = ["*"]
memory_read = ["*"]
memory_write = ["self.*"]