A charter linter gates the pilot branch on seven filled fields and a real revert command

The seven required fields are tested one at a time with `re.match` against every stripped non-empty line, so their order on the page does not matter, but each line has to begin with the field name ^[10][12]. Six of the patterns only check shape: an ISO date range, an @handle for each of the three roles, one non-empty token for scope and one for the stop rule ^[10].

Field seven, the revert path, is checked with a regex that stops at `git revert\b` ^[10]. The sample charter fills that field with `git revert -m 1 <merge-sha> on the pilot branch` ^[9]. That line passes. The first two words satisfy the pattern, and `<merge-sha>` is not one of the five substrings the placeholder check looks for ^[11][26]. The post writes that the revert path "must name a real command on a real branch" ^[19]; the script sees the first two words of the command ^[10].

A second pass lowercases the whole document and tests five substrings: tbd, todo, later, someone, and a bare question mark ^[11]. Any question mark anywhere on the page fails the run ^[28]. The scout, the charter says, "owns the question, not the merge" ^[2], and cannot punctuate one.

The rules that actually stop drift live outside the script. One-way handoffs, no self-approval by the scout, no scouting by the signer while a handoff sits open, no feature code from the signer, and one voice in shared chat: the linter tests none of these ^{[3][4][5][27]}.

Enforcement is a person reading notes. The stop rule is "two lost handoffs or one failed revert" ^[8], and the count comes from missing stamps read at the review ^[17]. That review is booked for 2026-09-19, three days into a window that opens 2026-09-16 and closes 2026-09-23 ^[6][24]. Stopping there leaves four days of the window unused ^[24].

Adoption is cheap in the literal sense: one wiki page, one script, no imports beyond `re`, `sys` and `pathlib` ^[22]. The failure output is one FAIL line per problem and exit status 1 ^[13]. The post says to wire that exit code into the same job that opens the pilot branch, and also to keep the linter "out of the pilot's critical path" ^[14][15]. Gating branch creation sits on the path to starting, so a team adopting this has to pick one of those two readings.

The post does not report a pilot count or an outcome before and after the charter; the handoff diagnosis is what the author says the logs usually show ^[29]. It ends on a pitch, with the claim that the first cost of a pilot "is rarely tokens" but the week spent provisioning seats and negotiating a budget line, followed by MonkeyCode's free model access and free server ^[20][21]. I would still run the page on an internal-service pilot. A named signer and a written revert command are worth the twenty minutes, and the script proves only that someone typed them ^[18][10].