The skill
log-messageas its source filebootstrap/skills/log-message.mdwrites it, rendered bygen-processes-viz.tsbecause it has no instruction page on this site — do not edit here. All processes
Logging what you are doing
Skill log-message: Say what you are doing, to the Logger, in a form a reader can act on. Five required fields and one optional body; an incomplete message is not logged. Callable from any task in any process, and required by some.
Named by: Complete initialization, Determine the harness and repositories, Initialize a harness, Log a message
You are the author of a log line; the Logger is where it goes. The Logger
records and decides nothing, so it cannot know who acted: you supply actor.
It is carried out as a sub-process: nobody starts there. A task in another process calls it, and the calling task’s diagram says so; this skill does not list its callers or the process that implements it.
The message
Five strings, all required, and one optional markdown body.
| field | what it answers | shape |
|---|---|---|
timestamp | when | an ISO-8601 instant, e.g. 2026-09-20T14:03:11Z |
actor | who was acting | the acting actor’s id — never the Logger’s |
process | inside which process | a process id, e.g. my-process |
task | at which step of it | a node id within that process, e.g. A_DoTheThing |
message | what happened | one line, no newline |
body | the detail, if any | markdown, optional |
All five are required: each answers a question a reader would otherwise guess.
body is the one place free prose belongs. A diff, an error, the contents
of a file that was not where it should have been. Keep message to the line
somebody scanning will read, and put everything else here.
When to log
Optionally, from any task in any process. You need no permission and no element in the diagram: an actor saying what it is doing is always in order.
Required where a diagram says so, which it says by drawing the call
explicitly — a callActivity whose calledElement is Process_LogMessage.
If the diagram drew it, logging is a step, not a courtesy.
The difference between the two is only whether the caller drew it. It is the same sub-process either way, and it takes the same six fields.
If you cannot complete the message
Do not log, and say that you could not. The End_NotLogged branch is a real
outcome, not an error path to route around.
A line missing its actor or its task still looks like a record, and a
reader cannot tell it from a complete one.
The same holds if the Logger is unreachable: say so, in the place you would have logged to, and carry on with the task. Logging is not a gate. A failure to log is never a reason to abandon an install.
The one destination bootstrap has
The discussion with the human actor. A Bootstrapping Agent has no log file, no service and no work plan — the only place it can put a line where a person will see it is the conversation it is already in.
There is no command: the agent writes the six fields into the conversation. A harness that installs a real log sink adds a Tool for it; this skill still says what a message is.
What this skill is NOT
A log line says what an actor did at a moment; it is not a record of work
still outstanding. Logging is also not asking: if a
step needs a judgement, that is confirm-harness, and no
number of log lines substitutes for it.