arminzou/dotbrain / plugin/skills/repo-rules
repo-rules
VerifiedInitializes and reconciles repository requirements in RULES.md and agent routing in AGENTS.md, and assesses enforcement gaps. Use when asked to establish repository rules, maintain adopted requirements, or assess or hand off their enforcement.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 7.1 KB |
| agents/openai.yaml | config | 118 B |
| references/rules-maintenance.md | markdown | 3.3 KB |
SKILL.md
View on GitHub---
name: repo-rules
description: Initializes and reconciles repository requirements in RULES.md and agent routing in AGENTS.md, and assesses enforcement gaps. Use when asked to establish repository rules, maintain adopted requirements, or assess or hand off their enforcement.
---
# Repository rules
Maintain adopted public repository requirements and their agent routing.
Requirements come from the repository's accepted guidance, configuration, and
authorized decisions. Dotbrain supplies no engineering baseline. When writing or
revising a rule, read the [rules maintenance examples](references/rules-maintenance.md)
for how to preserve authority, scope, and exceptions; their rules are illustrative.
## Select the operation
| Request | Operation |
| --- | --- |
| Initialize or reconcile repository rules. | Initialize or reconcile the repository's own requirements and routing below. Explicit initialization authorizes those documentation edits. |
| Maintain an adopted requirement. | Reconcile the scoped requirement with existing guidance; preserve unrelated content. |
| Assess coverage or enforcement. | Assess below, without editing files. Missing rules remain proposals. |
| Implement enforcement of a requirement. | Assess the scoped gap, then hand implementation to the selected engineering workflow. |
A bare invocation assesses the repository. Create or reconcile requirements only
when requested; invocation alone does not adopt proposed requirements. Apply user
and repository authority to every edit. Conflicting requirements need resolution;
continue independent authorized work without silently choosing a side.
## Initialize or reconcile
1. Identify the repository and read applicable instructions, existing requirements,
specialist guides, validation configuration, and CI. Inspect representative code
for actual conventions and dependency boundaries. Find the authoritative public
destination before editing; retain an existing standards or contributor guide
rather than introducing competing requirements.
2. Identify the source and scope of each requirement to document. Use accepted
guidance, explicit configuration, and authorized decisions; inspect code to
clarify application, not to turn incidental patterns into policy. If no adopted
requirements can be established, report the gap and propose candidates rather
than filling a rules file with generic practices. Invent no architecture,
commands, platforms, or adoption of new constraints.
3. Reconcile by meaning with the current content, rereading before saving if
another writer may have changed it. Preserve local amendments, existing
requirements, exceptions, rationale, and unrelated guidance. Account for each
scoped requirement as included, already covered, or conflicting, with its
adoption source. Improve existing application guidance when it already covers
the decision; add a rule only for a distinct adopted requirement. Keep proposals
and unresolved conflicts out of the adopted requirements.
4. Add or reuse a concise root `AGENTS.md` pointer directing agents making or
reviewing changes to read `RULES.md`, or the repository's existing authoritative
requirements destination. Create minimal routing if needed. Preserve aliases
and unrelated guidance; do not copy rules or README onboarding into routing.
5. Check coverage, local links, and named commands against actual configuration.
Review the complete diff for lost requirements or private context. Confirm that
repeating reconciliation would add no duplicate rules or routing and would
preserve local amendments.
Completion: the requested requirements and routing are reconciled, every scoped
requirement is accounted for, and conflicts remain explicitly unresolved.
## Rule format and authority
A rule states a durable requirement, its scope, and how contributors or reviewers
recognize compliance. A broad principle does not establish a concrete dependency
boundary or command; discover and name the accepted constraint. Personal
preferences and supported findings remain proposals until adopted. A recurring
failure alone does not authorize a new requirement or permanent workaround.
For a new `RULES.md`, use a concise format suited to the repository; the examples
show `Rule` / `How to apply` tables. Preserve an established requirements format
when reconciling. State each requirement clearly, keeping its scope, application,
observable checks, necessary rationale, exceptions, and links together. Escape
literal pipes when using table cells.
Keep each requirement in one authoritative location. Link specialist guides and
configuration for details instead of duplicating them, and check overlapping
guidance agrees. Public wording must stand alone in an ordinary
clone. Keep private evidence, personal paths, finding identifiers, and incident
history private when deriving an adopted public requirement.
## Assess enforcement
Read the scoped requirements, their adoption sources, affected code, existing
checks, and supplied failure examples. Assess coverage against the repository's
adopted requirements. Suggested additions remain proposals; absence of a generic
practice or an example rule is not a gap. If no requirements file exists,
recommend initialization rather than performing it.
For each scoped adopted rule, locate existing enforcement and identify the concrete
gap. Prefer an existing type or API boundary that prevents invalid choices, then
an existing static or architectural check, or a targeted behavioral test. Use a
mechanism that expresses the requirement; retain judgment-based decisions in
review guidance. Distinguish missing policy, missing enforcement, and a failure
to discover or follow guidance that already covers the decision.
Report the rule, adoption source, existing check location or command, observed
evidence, gap, smallest proposed change, and remaining limitation. Separate
inspection from demonstrated enforcement. Static evidence establishes static
constraints; runtime, integration, browser, and visual claims require evidence at
those levels. Local checks and observed CI runs are separate claims.
## Hand off enforcement
For requested enforcement, provide the selected engineering workflow with the
adopted rule and its authority, affected boundary and existing checks, proposed
mechanism, and verification needed to reject a representative violation and
accept valid behavior. Reuse an assessment only while it matches the repository.
Unadopted requirements and unresolved conflicts remain explicit decisions for
the appropriate authority before enforcement.
The selected workflow owns implementation in code, types, tests, lint, hooks, or
CI, along with verification and delivery procedures under user and repository
authority. This core operation does not implement those changes or prescribe
branching, delegation, review, pushes, PRs, or release steps. If no workflow is
selected, return the concrete assessment and handoff without claiming execution.
## Report
Identify changed files, requirement sources and coverage, unresolved conflicts,
checks actually performed, and remaining proposals. Distinguish written
requirements from demonstrated enforcement and work handed to the workflow.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |