REPO_POLICIES: the script/ entrypoints are the single source of truth; Makefile targets stay as shims #155

New Issue

2026-11-06T01:11:49+01:00

clawbot commented

2026-11-06 01:11:49 +01:00

sneak, 2026-11-06 in chat, on the rule "Always use Makefile targets (make fmt, make test, make lint, etc.) instead of invoking the underlying tools directly. The Makefile is the single source of truth for how these operations are run.": "we now use strta so calling those scripts is ok, they are the official single source of truth now. the makefile targets must still exist however"

Definition of done. In prompts/REPO_POLICIES.md on next, the rule at about line 202, and any checklist that restates it, says:

Run the repository's own entrypoints, never the underlying tools directly.
The script/ scripts (Scripts to Rule Them All) are the single source of truth for how each operation runs, and calling them directly is fine.
The Makefile targets must still exist, as the thin shims that call the scripts, as the document already describes.
The existing exception stands: a worker confined to one Go package may run go test on that package while iterating.
Plain short sentences; make fmt and make check pass.
One commit ending (closes #N), PR into next.

#154 edits the same file in a different section. Whichever lands second rebases.

Model: opus-5-5

sneak, 2026-11-06 in chat, on the rule "Always use Makefile targets (make fmt, make test, make lint, etc.) instead of invoking the underlying tools directly. The Makefile is the single source of truth for how these operations are run.": "we now use strta so calling those scripts is ok, they are the official single source of truth now. the makefile targets must still exist however" **Definition of done.** In `prompts/REPO_POLICIES.md` on `next`, the rule at about line 202, and any checklist that restates it, says: - Run the repository's own entrypoints, never the underlying tools directly. - The `script/` scripts (Scripts to Rule Them All) are the single source of truth for how each operation runs, and calling them directly is fine. - The Makefile targets must still exist, as the thin shims that call the scripts, as the document already describes. - The existing exception stands: a worker confined to one Go package may run `go test` on that package while iterating. - Plain short sentences; `make fmt` and `make check` pass. - One commit ending `(closes #N)`, PR into `next`. https://git.eeqj.de/sneak/prompts/issues/154 edits the same file in a different section. Whichever lands second rebases. Model: opus-5-5

clawbot referenced this issue from a commit

2026-11-06 01:12:44 +01:00

REPO_POLICIES: script/ entrypoints are the source of truth (closes #155)

clawbot referenced a pull request that will close this issue

2026-11-06 01:12:52 +01:00

REPO_POLICIES: script/ entrypoints are the single source of truth #157

clawbot commented

2026-11-06 01:12:56 +01:00

Rewrote the rule in REPO_POLICIES.md: run the repo's own entrypoints, the script/ scripts are the single source of truth and may be called directly, and the Makefile targets must still exist as thin shims. The one-package go test exception stands. PR: #157

Model: opus-5-5

Rewrote the rule in `REPO_POLICIES.md`: run the repo's own entrypoints, the `script/` scripts are the single source of truth and may be called directly, and the Makefile targets must still exist as thin shims. The one-package `go test` exception stands. PR: https://git.eeqj.de/sneak/prompts/pulls/157 Model: opus-5-5

clawbot referenced this issue

2026-11-06 01:18:27 +01:00

Compact REPO_POLICIES.md without losing fidelity or clarity #158

clawbot closed this issue

2026-11-06 01:18:53 +01:00

clawbot referenced this issue

2026-11-06 01:58:42 +01:00

Compact REPO_POLICIES.md without losing rules (closes #158) #159

Sign in to join this conversation.