graduated-validation-protocol

graduated-validation-protocol is a skill for Claude Code, Codex from AnthonyAlcaraz/agentic-graph-rag-skills. It costs 216 tokens per session (3,426 once invoked), scanned A, original, MIT.

A safety process for testing changes made by a self-improving software agent before they reach production. It assigns each change a risk level and applies a matching review or release test.

In plain words
What is it for?
Use it to validate candidate changes through canary releases, staging, or stricter checks according to their risk.
Why use it?
It reduces the chance that an automatically generated change will harm live users and keeps changes traceable and reversible.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit Use it to validate candidate changes through canary releases, staging, or stricter checks according to their risk.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/anthonyalcaraz/agentic-graph-rag-skills/graduated-validation-protocol
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add AnthonyAlcaraz/agentic-graph-rag-skills --skill graduated-validation-protocol
Clone the repo
git clone --depth 1 https://github.com/AnthonyAlcaraz/agentic-graph-rag-skills

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for graduated-validation-protocol

README.md
[![agentmods](https://agentmods.dev/badge/skills/anthonyalcaraz/agentic-graph-rag-skills/graduated-validation-protocol/github.svg)](https://agentmods.dev/skills/anthonyalcaraz/agentic-graph-rag-skills/graduated-validation-protocol)
Your own site
<a href="https://agentmods.dev/skills/anthonyalcaraz/agentic-graph-rag-skills/graduated-validation-protocol"><img src="https://agentmods.dev/badge/skills/anthonyalcaraz/agentic-graph-rag-skills/graduated-validation-protocol/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for graduated-validation-protocol

Your own site · 80×15
<a href="https://agentmods.dev/skills/anthonyalcaraz/agentic-graph-rag-skills/graduated-validation-protocol"><img src="https://agentmods.dev/badge/skills/anthonyalcaraz/agentic-graph-rag-skills/graduated-validation-protocol.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 216 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,426 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00216 $0.03426
Opus 5 $0.00108 $0.01713
Sonnet 5 $0.00043 $0.00685
Haiku 4.5 $0.00022 $0.00343

Measured 10d ago against content hash 09a5b2a5763f, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade A, and why

graduated-validation-protocol scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.

The scan reads SKILL.md. This mod also ships 2 executable files (cli.py, lib.py), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/self-evolution/graduated-validation-protocol/SKILL.md · 211 lines

How it starts

The opening of the file, as written. The whole thing — 211 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Graduated Validation Protocol

Overview

The evolutionary loop generates candidate improvements continuously. Not all of them should reach production. Any system that modifies itself in production needs a safety envelope. Ch7 provides it through the RPO spine and the Graduated Validation Protocol (GVP).

The RPO spine is three interlocking properties:

  • Recursion — the improvement mechanisms are themselves part of the codebase the agent can analyze and improve. The self-improvement system applies to itself.
  • Provenance — every change is cryptographically signed and committed to an immutable ledger. This is source control for thought: complete auditability and instant rollback for every model, prompt, or architectural change. lib.provenance_signature is the dev-time stand-in.
  • Optimization — the GVP itself. Rather than a single approval gate that must be either permissive or strict, it routes each candidate to a risk tier and applies matching scrutiny.

The three tiers:

Tier Risk class Gate Outcome
1 canary release low: minor prompt updates, hyperparameter adjustments ~1% live traffic; statistically significant target-metric lift with no core-KPI degradation pass or automatic rollback
2 staging gauntlet moderate: SEAL/TPT fine-tune results, refined tool integration benchmark + regression (catastrophic forgetting) + safety/alignment (bias drift) + performance (latency/cost) suites; net-positive multi-objective utility U = w_accuracyaccuracy + w_cost(1-cost) + w_safety*safety_score, no safety regressions pass or archive
3 airlock protocol high: modifies safety protocols, integrates a brand-new tool, or failed staging with large potential sandboxed isolated environment; automated risk/reward report escalated for human approve / reject / modify

The second half is the entropy-collapse guard. In production the agent evolves continuously, hundreds of small improvements per week, each passing validation individually. The compounding problem is the accumulated weight of all changes: contradictory learnings, stale fixes for resolved issues, redundant nodes that slow retrieval. The Kepler architecture (OpenAI, 2026) solves this with a dual-store pattern: Knowledge is human-authored and version-controlled; Learnings are agent-generated and ephemeral. A daily garbage-collection traversal walks the Learnings subgraph and removes a node if any of three conditions holds: the issue was resolved and promoted to Knowledge, a newer higher-confidence learning contradicts it, or it has not been retrieved in 30 days.

Read the full file on GitHub · 211 lines

Files

What ships with it

2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 10d ago First seen · 211 lines · 216 tokens per session scan A 09a5b2a5763f

Subscribe to this mod's changes

graduated-validation-protocol is a skill published in the GitHub repository AnthonyAlcaraz/agentic-graph-rag-skills (10 stars, last pushed 1mo ago), licensed MIT. It adds 216 tokens to every session and 3,426 once invoked, about $0.0011 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

foundry-hosted-agent-validation

Step-by-step process for validating a Python Foundry hosted agent sample (under python/samples/04-hosting/foundry-hosted-agents/) end to end — running it locally (native runtime and azd ai agent run) and after deploying it to an Azure AI Foundry project with azd. Use this when asked to validate a hosted agent sample.

microsoft/agent-framework · 82 tokens

build-and-test

How to build and test .NET projects in the Agent Framework repository. Use this when verifying or testing changes.

microsoft/agent-framework · 26 tokens

workflow-patterns

Use this skill when implementing tasks according to Conductor's TDD workflow, handling phase checkpoints, managing git commits for tasks, or understanding the verification protocol.

wshobson/agents · 35 tokens

ios-simulator

Verify and debug native, React Native, Expo, or Flutter apps on an iOS Simulator with agent-device. Use when an agent needs to launch an app, inspect its live UI, tap, type, scroll, validate a code change, collect failure evidence, or reproduce a workflow on an iPhone or iPad Simulator.

callstack/agent-device · 69 tokens

agent-evaluation

Run one specified Test Agent on one specified Benchmark Case exactly once, privately score that execution, and return one protocol result.

Prism-Shadow/penguin-harness · 28 tokens

jetson-validate-image

Use after jetson-flash-image to run static BSP checks, on-target smoke/regression tests on a flashed DUT, or both. Not for build or flash steps. Triggers: validate bsp, on-target validation.

NVIDIA/skills · 50 tokens