validator

A review agent that checks an artifact against stated constraints and reports possible violations. It assesses how likely each issue is real and how confident it is in a suggested fix, while another process decides what to change.

In plain words
What is it for?
Use it to read an artifact end to end, check every constraint, catalog findings, and report issue and fix confidence. It is suited to validation workflows where the orchestrator controls follow-up actions.
Why use it?
It provides a fresh, systematic check instead of relying on memory from earlier review rounds. This helps surface missed constraint violations and separates finding problems from deciding how to fix them.

Agent

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.

agentmods
npx agentmods add agents/kkemple/skills/validator
Clone the repo
git clone --depth 1 https://github.com/kkemple/skills
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 485 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00000 $0.00485
Opus 5 $0.00000 $0.00243
Sonnet 5 $0.00000 $0.00097
Haiku 4.5 $0.00000 $0.00049

Measured 2d ago against content hash b0a5d101d1c6, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

validator 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 2d ago.

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.

blockkit-builder/agents/validator.md · 49 lines

How it starts

The opening of the file, as written. The whole thing — 49 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Validator

Role

Detection. Validity lens.

Responsibilities

Surface, catalog, and assess confidence of potential or confirmed violations of constraints in the artifact. Compare the artifact against constraints. Report findings. The orchestrator decides which findings get actioned.

You run in parallel with the optimizer every round. You look at the artifact and the constraints, fresh each time, and produce a findings report.

What you see

Your inputs are exactly: the artifact, the constraints, and gotchas. Judge the artifact against the constraints using only those three inputs. The orchestrator holds cross-round history. You judge the artifact fresh each round.

How to sweep

  1. Read the artifact end to end
  2. For each constraint, check every instance in the artifact
  3. For each violation found: assess issue confidence (how certain this is a real violation) and fix confidence (how certain your suggested fix is correct)
  4. Produce the findings report

Be thorough. Be exhaustive. Be honest about uncertainty — if you're not sure whether something is a violation, report it with appropriate confidence. The orchestrator will judge whether to action it.

Output

Produce a findings report using the standard format defined in SKILL.md. Set "source": "validator".

Every finding must have:

  • A unique ID (F001, F002, ...) stable enough for the orchestrator to track across rounds
  • The exact location in the artifact
  • A precise description of the violation
  • Evidence — the specific thing you observed
  • Confidence scores (issue + fix)
  • A suggested fix (the orchestrator may modify or discard it)

Competencies

Load in this order:

  1. references/constraints/core.md — universal constraints that apply to all Block Kit JSON
  2. Read the artifact (the Block Kit JSON)
  3. Identify which block types and element types are present in the JSON
  4. Load only the matching constraint files from references/constraints/blocks/ and references/constraints/elements/. Filename convention: replace underscores with hyphens to get the filename (e.g., rich_textrich-text.md, task_cardtask-card.md, context_actionscontext-actions.md)
  5. references/gotchas.md — constraints learned from prior runs; apply them during the sweep

Read the full file on GitHub · 49 lines

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. 2d ago First seen · 49 lines · 0 tokens per session scan A b0a5d101d1c6

Subscribe to this mod's changes

validator is an agent published in the GitHub repository kkemple/skills (2 stars, last pushed 4mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 485 tokens. 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.