lN-review-doctrine
Skill4 starsChanged 3 months agoMITView on GitHub
Shared doctrine reference for the Acceptance and Objective review skills (l1-review, l2-review). Holds the 3-axis checklist (Conformance, Process, Objective Advancement), the reviewer-independence and reviewer-worktree-discipline invariants, the work-type → required-verification map consumed by axis 2, the finding/verdict schema, and the self-PR posting caveat. This skill has no operations — it is reference content read by `/l1-review` and `/l2-review` at review time. Never inline this doctrine into the l1-review or l2-review skills themselves — operator-confirmed rules must persist in one place.
In this repository41 other files it gives its agents.
- cadence-goalsSkill
- ci-feedback-loopSkill
- compoundSkill
- decision-rights-doctrineSkill
- diataxisSkill
- dispatch-gateSkill
- drift-checkSkill
- family-checkSkill
- fix-then-re-review-ladderSkill
- github-projectsSkill
- gitSkill
- goal-treeSkill
- handoffSkill
- improvement-loopSkill
- kb-captureSkill
- kb-compileSkill
- kb-coreSkill
- kb-lessons-bridgeSkill
- kb-lintSkill
- l1-supervisorSkill
- l2-supervisorSkill
- lessons-learnedSkill
- lN-lifecycle-doctrineSkill
- medical-recordsSkill
- operator-interview-doctrineSkill
- operator-interviewSkill
- operator-reviewSkill
- planning-workflowSkill
- pr-review-fanoutSkill
- pull-taskSkill
- ralph-wiggumSkill
- reviewSkill
- root-cause-depthSkill
- scan-documentSkill
- self-improvementSkill
- self-verifySkill
- skillifySkill
- task-workflowSkill
- testing-without-mocksSkill
- vault-cloudSkill
- verify-inherited-premiseSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 54
If a rule in references/ is wrong for the current iteration,
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
lN-review-doctrine — shared Acceptance/Objective review doctrine
This skill is a doctrine-only reference. It defines what an
Acceptance or Objective review is, regardless of which layer runs
it. The actual review skills (l1-review, l2-review — the
Acceptance and Objective review executors) are thin executors that
load this doctrine at review time and apply it at their layer.
Source of truth
| File | Purpose |
references/checklist.md | Role, reviewer-independence invariant, reviewer-worktree-discipline invariant, the 3-axis checklist, finding schema, verdict rubric (incl. the single-vocabulary deprecation of APPROVE/REJECT/ESCALATE), artifact format, posting protocol, local-artifact rule. |
references/verification-map.md | Work-type → required-verification map consumed by axis 2 (Process), plus the doctrine- and workflow-class sub-checklists. |
references/depth-rule.md | Depth/proportionality rule — the change-class table, the load-bearing surfaces list, and the overridable size-threshold parameters that determine how much of the checklist applies to a given work product. |
references/dispatch-procedure.md | Event-triggered review dispatch — the depth-rule-gated chain fired from PR creation (instead of waiting for a supervisor tick to notice pr-open). Consumed by gh-pr-create.md (dotfiles) and task-workflow/operations/finish.md. |
Invariants
An Acceptance or Objective review never re-runs /review.
Code-level correctness, security, simplicity, and architecture
review is the Change review's (L0's /review) job; the
Acceptance/Objective review skill consumes the verdict the Change
review posts to the PR (the <!-- review:metadata --> marker — the
cross-operator artifact, not the gitignored local
.claude/reviews/latest.md) as evidence for axis 2.
An Acceptance or Objective review always evaluates against the reviewing layer's parent objective, not the reviewed layer's task description. This is the recursive lever; see checklist.md axis 3 for what to read to determine the reviewing layer's objective.
An Acceptance or Objective review runs in an independent context window — a fresh verifier that sees only the work product plus the reviewing layer's parent objective, never the authoring trace. See checklist.md Reviewer independence.
An Acceptance or Objective review operates read-only on the
worktree — never stash, checkout, reset, restore, clean,
rebase, or switch branches in the review worktree, even
temporarily. See checklist.md Reviewer worktree discipline for
the non-mutating recipe for inspecting other revisions.
When this doctrine is wrong
If a rule in references/ is wrong for the current iteration,
fix it here via PR against pfeff/claude-skills. Do not patch
the rule inside l1-review or l2-review — that's the
doctrine-drift failure mode this split exists to prevent.
See also
lN-lifecycle-doctrine— the parallel lifecycle doctrine; itspr-open → fixing → mergedstates drive when an Acceptance or Objective review runs.l1-reviewskill — executes this doctrine as the Acceptance review (N=1).l2-reviewskill — executes this doctrine as the Objective review (N=2).mbp/review(plugin) — the Change review's (L0's) per-line review skill; its output is read by axis 2 of an Acceptance or Objective review. Out of scope for changes here.references/dispatch-procedure.md— how the chain gets triggered in the first place: dispatched from PR creation rather than only from a supervisor tick or a manual invocation.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

