validate

A validation workflow for checking whether the Cobalt project matches WhatsApp Web, Desktop, and Mobile behavior. It compares both the source-level implementation and the observable events, data nodes, and HTTP requests it produces.

In plain words
What is it for?
Use it to validate WhatsApp-related features, inspect their produced events and requests, preserve validation results, and identify mismatches with the real WhatsApp runtime.
Why use it?
It provides a structured way to prove behavioral compatibility instead of relying only on compilation or manual inspection. It also enforces project-specific checks before validation runs.

Command for Claude Code

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 commands/auties00/cobalt/validate
Clone the repo
git clone --depth 1 https://github.com/Auties00/Cobalt

Made for: Claude Code.

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 10,206 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 $0.10206
Opus 5 $0.00000 $0.05103
Sonnet 5 $0.00000 $0.02041
Haiku 4.5 $0.00000 $0.01021

Measured 2d ago against content hash 13922e5a3206, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

validate 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.

.claude/commands/validate.md · 750 lines

How it starts

The opening of the file, as written. The whole thing — 750 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Cobalt Validation Orchestrator

You are the lead validation orchestrator for the Cobalt project. Your job is to prove that Cobalt implements every WhatsApp Web / Desktop / Mobile feature correctly, both at the source level (behavioral parity with the WA Web JS source) and at the observable level (the Nodes, WAM events and HTTP that Cobalt produces for a given input must match what the real WhatsApp runtime produces for the same input).

The user invokes this command as: /validate.

Preconditions

  • Work from the Cobalt repository root.
  • Preserve existing validation outputs under validation/ across runs.
  • NEVER run mvn yourself. The user uses their IDE for compile diagnostics. Agents never run mvn compile either — scratch-file compilation is the one exception the agent is allowed, and only via the IDE-agnostic mechanism documented in validate-module.

Clean Working Tree (required)

/validate requires a clean working tree. Run git status --porcelain; if it prints anything, STOP and tell the user to commit (or otherwise clear) their changes first. This is a hard precondition, not a soft warning, and there is no dirty-tree fallback.

The clean-tree rule is what makes incremental mode exact. A run starts from a known commit and ends (Phase 4.5) by committing the fixes its agents applied, so the next run can diff those two commits with neither false positives nor missed changes. A dirty tree at either boundary would smear that delta.

Verify MCP Server

Before doing anything else, verify the whatsapp MCP server is reachable:

  1. Call any lightweight MCP tool such as mcp__whatsapp__get_active_snapshot.
  2. If the server is NOT running, ask the user to start it. This is a hard precondition, not a soft warning.

Determine Run Mode (incremental vs full)

Decide FULL or INCREMENTAL once, here, and record it as the first line of scope.md.

Run INCREMENTAL when ALL of these hold:

  • validation/run-state.json exists and its schemaVersion equals the current schema (1).
  • Every snapshots.<platform>.snapshotId it records still appears in list_snapshots(<platform>), so get_revision_diff can resolve its from.
  • Its cobaltCommit is reachable: git cat-file -e <cobaltCommit>^{commit} succeeds.

Read the full file on GitHub · 750 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 · 750 lines · 0 tokens per session scan A 13922e5a3206

Subscribe to this mod's changes

validate is a command published in the GitHub repository Auties00/Cobalt (915 stars, last pushed 1mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 10,206 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 2026-08-30.