agentsclimarketplace

Governance review

Skill canstralian/rif-runtime/.claude/skills/governance-review

Review capabilities, providers, execution paths, and policy enforcement to ensure every effectful operation remains governed, auditable, and replayable.From its SKILL.md

Install
npx -y skills add canstralian/rif-runtime --skill governance-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.4 KB, as published. Nobody here has run it

Governance Review

Mission

Protect the Governance Plane.

Every effectful operation performed by the runtime must pass through explicit governance before execution.

No execution path may bypass policy evaluation.


Responsibilities

Review:

  • Capability descriptors
  • Policy decisions
  • Provider integrations
  • Execution plans
  • Effect classification
  • Evidence generation
  • Replay compatibility

Governance Pipeline

Intent

↓

Planner

↓

Capability Resolution

↓

Policy Evaluation

↓

Execution

↓

Evidence

↓

Replay

↓

Receipts

Every execution must follow this sequence.


Required Capability Properties

Every capability must define:

  • unique name
  • resource kind
  • effect type
  • replayable status
  • description

Capabilities must be deterministic.


Policy Rules

Policy evaluates.

Providers execute.

Execution never performs policy decisions.

Policy decisions must be explicit.


Effect Classification

Every operation must declare an effect.

Supported effects include:

  • READ
  • WRITE
  • SNAPSHOT
  • INVENTORY
  • PROJECT
  • RENDER

Adding a new effect requires architectural review.


Provider Review

Providers:

  • perform work
  • return results
  • emit receipts

Providers never:

  • own policy
  • mutate evidence
  • bypass governance

Evidence Requirements

Every governed effect records:

  • execution identifier
  • capability
  • provider
  • resource
  • timestamp
  • policy decision
  • outcome

Evidence is immutable.


Replay Requirements

Replay must reproduce:

  • capability
  • policy decision
  • inputs
  • outputs
  • receipts

Replay must never modify historical evidence.


Review Checklist

Reject changes that:

  • bypass Governance
  • perform writes without policy
  • hide provider behaviour
  • omit evidence
  • weaken replay
  • duplicate capability names
  • introduce implicit permissions

Required Outputs

Produce:

Governance Summary

Capability Review

Policy Review

Evidence Review

Replay Review

Risk Assessment


Success Criteria

Approve only implementations that preserve explicit governance, deterministic execution, immutable evidence, and replayability.

Every governed effect must be explainable, auditable, and reproducible.

Keep looking

Skills are one crate of 608,666. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.