| name | chat |
| description | Post a short (max 255 chars) message to the team chat log (agents/CHAT.md). Use to communicate between personas, log progress updates, and coordinate handoffs between agents. |
| triggers | ["*chat","*msg","*chat log"] |
Chat Skill
Overview
The chat skill posts structured messages to agents/CHAT.md, the shared team communication log. All personas use this to coordinate work and hand off tasks.
Usage
make chat MSG="<message>" [PERSONA="<Name>"] [CMD="<command>"] [TO="<recipient>"]
Arguments
| Argument | Variable | Default | Description |
|---|
| message | MSG | required | Message content |
| persona | PERSONA | $USER | Who is sending (e.g. Neo, Trin) |
| cmd | CMD | chat | Command prefix (auto-prefixed with *) |
| to | TO | all | Recipient persona name |
Output Format
[DATETIME] [**Persona**]->[**recipient**] *cmd*:
message
Examples
Log a user request
make chat MSG="fix the bug in parser.py" PERSONA="User" CMD="request"
Post a persona response
make chat MSG="Fixed bug in parser.py line 42" PERSONA="Neo" CMD="swe fix" TO="Trin"
Assign work to another persona
make chat MSG="@Trin please verify the fix in parser.py" PERSONA="Neo" CMD="handoff" TO="Trin"
The 255-char limit
make chat hard-fails (no partial post, no truncation) if MSG exceeds 255 characters. There is
no proactive warning while drafting — the failure only surfaces after the call. Before writing a
long message: count first, or default to the fallback below rather than discovering the limit
by hitting it.
Fallback pattern for anything long (detailed findings, UAT reports, multi-file summaries):
- Write the full content to a doc file:
agents/<persona>.docs/<TOPIC>_<YYYY-MM-DD>.md.
- Post a short pointer instead:
make chat MSG="<one-line summary>. Details: <path>" ....
This is the same pattern already used throughout agents/*.docs/ for UAT reports and sprint
summaries — apply it to the chat message itself whenever the content won't fit, not just to
"detailed technical notes" after the fact.
When to Post
- ENTRY: After reading CHAT.md to acknowledge context
- WORK: After completing each significant step
- HANDOFF: When switching to another persona — assign the next task explicitly
- HELP: When your are not sure what to do next and need help from another agent or human
- EXIT: Before saving state files
Reading the Chat Log
Always read agents/CHAT.md (newest messages at the END) before starting work:
Read agents/CHAT.md # last 10-20 messages for context
One-line summary: Posts structured messages to the shared team chat log at agents/CHAT.md.
TLDR:
Use make chat MSG="..." PERSONA="..." CMD="..." TO="..." to log persona activity and coordinate handoffs between agents.
All personas should post on entry, after significant work steps, on handoff, and before saving state on exit.
Newest messages are at the END of agents/CHAT.md — always read the bottom for current context.
