From 2778010397735cb12dfdd21bcf569bdb6456239b Mon Sep 17 00:00:00 2001 From: Pieter Viljoen Date: Fri, 28 Aug 2026 20:16:03 -0700 Subject: [PATCH 1/2] Drop the invented Close: field, match AGENTS.md's own unlabeled shape (#1070 follow-up) Copilot's second promotion-PR round flagged the Close: field as a non-canonical key: AGENTS.md's own template has 5 labeled fields followed by one unlabeled trailing sentence, not 6 labeled fields. Matches that shape exactly now while keeping the 'quote X into the prompt' phrasing that has drawn no ambiguity complaint across 4 rounds. --- .agents/skills/local-strict-review/SKILL.md | 4 ++-- .claude-plugin/fleet-skills/.source-digest | 2 +- .../fleet-skills/skills/local-strict-review/SKILL.md | 4 ++-- .github/skills/local-strict-review/SKILL.md | 4 ++-- 4 files changed, 7 insertions(+), 7 deletions(-) diff --git a/.agents/skills/local-strict-review/SKILL.md b/.agents/skills/local-strict-review/SKILL.md index 27f1238e..89bc7595 100644 --- a/.agents/skills/local-strict-review/SKILL.md +++ b/.agents/skills/local-strict-review/SKILL.md @@ -48,8 +48,8 @@ Rules that bind this task: quote `code-review`'s "Review the Change" section int the Return line below, not a PR comment or a coverage marker. Return: one finding per line, file:line, the concrete failure scenario, no severity theater. Bounds: read-only. No edit, no stage, no commit, no push, no PR-hosted write of any kind. -Close: quote `AGENTS.md`'s own unresolved-rule closing line ("Context and Delegation - Discipline") into the prompt, do not restate it here. +Quote `AGENTS.md`'s own unresolved-rule closing line ("Context and Delegation Discipline") + into the prompt as this brief's own final line, do not restate it here. ``` **Model tier:** the strongest tier this session can reach, per `AGENTS.md` "Match the model tier to the judgment" and "Never tier down the seat holding the judgment", applied here to the reviewer rather than the author. Run the pass on the same tier that authored the change when only one tier is reachable, a second, adversarially-prompted look still catches what the authoring pass's own "looks ready" judgment did not. diff --git a/.claude-plugin/fleet-skills/.source-digest b/.claude-plugin/fleet-skills/.source-digest index f470fe62..774af6b4 100644 --- a/.claude-plugin/fleet-skills/.source-digest +++ b/.claude-plugin/fleet-skills/.source-digest @@ -1 +1 @@ -4dbdf71167625f52 +f06e670f698e2a94 diff --git a/.claude-plugin/fleet-skills/skills/local-strict-review/SKILL.md b/.claude-plugin/fleet-skills/skills/local-strict-review/SKILL.md index 27f1238e..89bc7595 100644 --- a/.claude-plugin/fleet-skills/skills/local-strict-review/SKILL.md +++ b/.claude-plugin/fleet-skills/skills/local-strict-review/SKILL.md @@ -48,8 +48,8 @@ Rules that bind this task: quote `code-review`'s "Review the Change" section int the Return line below, not a PR comment or a coverage marker. Return: one finding per line, file:line, the concrete failure scenario, no severity theater. Bounds: read-only. No edit, no stage, no commit, no push, no PR-hosted write of any kind. -Close: quote `AGENTS.md`'s own unresolved-rule closing line ("Context and Delegation - Discipline") into the prompt, do not restate it here. +Quote `AGENTS.md`'s own unresolved-rule closing line ("Context and Delegation Discipline") + into the prompt as this brief's own final line, do not restate it here. ``` **Model tier:** the strongest tier this session can reach, per `AGENTS.md` "Match the model tier to the judgment" and "Never tier down the seat holding the judgment", applied here to the reviewer rather than the author. Run the pass on the same tier that authored the change when only one tier is reachable, a second, adversarially-prompted look still catches what the authoring pass's own "looks ready" judgment did not. diff --git a/.github/skills/local-strict-review/SKILL.md b/.github/skills/local-strict-review/SKILL.md index 27f1238e..89bc7595 100644 --- a/.github/skills/local-strict-review/SKILL.md +++ b/.github/skills/local-strict-review/SKILL.md @@ -48,8 +48,8 @@ Rules that bind this task: quote `code-review`'s "Review the Change" section int the Return line below, not a PR comment or a coverage marker. Return: one finding per line, file:line, the concrete failure scenario, no severity theater. Bounds: read-only. No edit, no stage, no commit, no push, no PR-hosted write of any kind. -Close: quote `AGENTS.md`'s own unresolved-rule closing line ("Context and Delegation - Discipline") into the prompt, do not restate it here. +Quote `AGENTS.md`'s own unresolved-rule closing line ("Context and Delegation Discipline") + into the prompt as this brief's own final line, do not restate it here. ``` **Model tier:** the strongest tier this session can reach, per `AGENTS.md` "Match the model tier to the judgment" and "Never tier down the seat holding the judgment", applied here to the reviewer rather than the author. Run the pass on the same tier that authored the change when only one tier is reachable, a second, adversarially-prompted look still catches what the authoring pass's own "looks ready" judgment did not. From 2faa0e0663932e0abced23fd4f753fde349ef2eb Mon Sep 17 00:00:00 2001 From: Pieter Viljoen Date: Fri, 28 Aug 2026 20:37:31 -0700 Subject: [PATCH 2/2] Use an explicit bracket placeholder, matching AGENTS.md's own template syntax (#1070 follow-up) Copilot's fourth round asked for exactly this: the unlabeled imperative sentence still read as an instruction addressed to the subagent, who cannot fulfil it (no AGENTS.md access). AGENTS.md's own template already uses <...> placeholder syntax for every field value; this is the second use of that same convention in this file (the first is at line 40), not a new pattern, and fails safe if pasted unsubstituted (an obvious unfilled token, not a misleading real instruction). --- .agents/skills/local-strict-review/SKILL.md | 4 ++-- .claude-plugin/fleet-skills/.source-digest | 2 +- .../fleet-skills/skills/local-strict-review/SKILL.md | 4 ++-- .github/skills/local-strict-review/SKILL.md | 4 ++-- 4 files changed, 7 insertions(+), 7 deletions(-) diff --git a/.agents/skills/local-strict-review/SKILL.md b/.agents/skills/local-strict-review/SKILL.md index 89bc7595..2349271d 100644 --- a/.agents/skills/local-strict-review/SKILL.md +++ b/.agents/skills/local-strict-review/SKILL.md @@ -48,8 +48,8 @@ Rules that bind this task: quote `code-review`'s "Review the Change" section int the Return line below, not a PR comment or a coverage marker. Return: one finding per line, file:line, the concrete failure scenario, no severity theater. Bounds: read-only. No edit, no stage, no commit, no push, no PR-hosted write of any kind. -Quote `AGENTS.md`'s own unresolved-rule closing line ("Context and Delegation Discipline") - into the prompt as this brief's own final line, do not restate it here. + ``` **Model tier:** the strongest tier this session can reach, per `AGENTS.md` "Match the model tier to the judgment" and "Never tier down the seat holding the judgment", applied here to the reviewer rather than the author. Run the pass on the same tier that authored the change when only one tier is reachable, a second, adversarially-prompted look still catches what the authoring pass's own "looks ready" judgment did not. diff --git a/.claude-plugin/fleet-skills/.source-digest b/.claude-plugin/fleet-skills/.source-digest index 774af6b4..39a139f0 100644 --- a/.claude-plugin/fleet-skills/.source-digest +++ b/.claude-plugin/fleet-skills/.source-digest @@ -1 +1 @@ -f06e670f698e2a94 +2044c888625ec970 diff --git a/.claude-plugin/fleet-skills/skills/local-strict-review/SKILL.md b/.claude-plugin/fleet-skills/skills/local-strict-review/SKILL.md index 89bc7595..2349271d 100644 --- a/.claude-plugin/fleet-skills/skills/local-strict-review/SKILL.md +++ b/.claude-plugin/fleet-skills/skills/local-strict-review/SKILL.md @@ -48,8 +48,8 @@ Rules that bind this task: quote `code-review`'s "Review the Change" section int the Return line below, not a PR comment or a coverage marker. Return: one finding per line, file:line, the concrete failure scenario, no severity theater. Bounds: read-only. No edit, no stage, no commit, no push, no PR-hosted write of any kind. -Quote `AGENTS.md`'s own unresolved-rule closing line ("Context and Delegation Discipline") - into the prompt as this brief's own final line, do not restate it here. + ``` **Model tier:** the strongest tier this session can reach, per `AGENTS.md` "Match the model tier to the judgment" and "Never tier down the seat holding the judgment", applied here to the reviewer rather than the author. Run the pass on the same tier that authored the change when only one tier is reachable, a second, adversarially-prompted look still catches what the authoring pass's own "looks ready" judgment did not. diff --git a/.github/skills/local-strict-review/SKILL.md b/.github/skills/local-strict-review/SKILL.md index 89bc7595..2349271d 100644 --- a/.github/skills/local-strict-review/SKILL.md +++ b/.github/skills/local-strict-review/SKILL.md @@ -48,8 +48,8 @@ Rules that bind this task: quote `code-review`'s "Review the Change" section int the Return line below, not a PR comment or a coverage marker. Return: one finding per line, file:line, the concrete failure scenario, no severity theater. Bounds: read-only. No edit, no stage, no commit, no push, no PR-hosted write of any kind. -Quote `AGENTS.md`'s own unresolved-rule closing line ("Context and Delegation Discipline") - into the prompt as this brief's own final line, do not restate it here. + ``` **Model tier:** the strongest tier this session can reach, per `AGENTS.md` "Match the model tier to the judgment" and "Never tier down the seat holding the judgment", applied here to the reviewer rather than the author. Run the pass on the same tier that authored the change when only one tier is reachable, a second, adversarially-prompted look still catches what the authoring pass's own "looks ready" judgment did not.