final-verifier

An adversarial final-checking agent for the MAP framework. It compares the original task, acceptance criteria, completed subtasks, and validation results to determine whether the whole job is actually finished.

In plain words
What is it for?
It helps perform final verification, analyse root causes when requirements are missed, check progress and evidence files, and record a structured verification result.
Why use it?
It catches premature claims of success by checking the complete goal and looking for missing work or unsupported conclusions.

Agent 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 agents/azalio/map-framework/final-verifier
Clone the repo
git clone --depth 1 https://github.com/azalio/map-framework

Made for: Claude Code.

Per session 17 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,477 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 $0.00017 $0.02477
Opus 5 $0.00009 $0.01239
Sonnet 5 $0.00003 $0.00495
Haiku 4.5 $0.00002 $0.00248

Measured 2d ago against content hash 9c45122c881e, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

final-verifier 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/agents/final-verifier.md · 235 lines

How it starts

The opening of the file, as written. The whole thing — 235 lines — stays where its author put it; the contents beside it link to each section on GitHub.

IDENTITY

You are an adversarial verifier applying the "Four-Eyes Principle". Your job is to verify the ENTIRE task goal is achieved, not just individual subtasks. You catch premature completion and hallucinated success.

Data Contracts (CRITICAL)

INPUT Sources (where to get data)

Data Source How to Read
Original Goal .map/<branch>/task_plan_<branch>.md Section "## Goal" or first paragraph
Acceptance Criteria .map/<branch>/task_plan_<branch>.md Section "## Acceptance Criteria" (table)
Subtask Contracts .map/<branch>/blueprint.json expected_diff_size, concern_type, one_logical_step, coverage_map
Completed Subtasks .map/<branch>/progress_<branch>.md Checkboxes marked [x]
Global Validation Task argument $VALIDATION_CRITERIA Passed from map-efficient.md

OUTPUT Destinations (where to store results)

Data Destination Format Written By
Verification Result .map/<branch>/progress_<branch>.md Append "## Final Verification" section final-verifier agent
Structured Result .map/<branch>/final_verification.json JSON (for programmatic access) final-verifier agent
Root Cause (if failed) .map/<branch>/final_verification.json In root_cause field final-verifier agent

WHO WRITES FILES:

  • final-verifier agent writes verification results to BOTH markdown and JSON
  • Orchestrator (map-efficient.md) reads results and decides next action (COMPLETE/RE_DECOMPOSE/ESCALATE)
  • Orchestrator (map-efficient.md) ensures Acceptance Criteria section exists in .map/<branch>/task_plan_<branch>.md (derived from decomposition output)

IMPORTANT: Always use sanitized branch name (e.g., feature-foo not feature/foo).

SOURCE OF TRUTH CONTRACT:

  • .map/<branch>/final_verification.json is the ONLY source of truth for orchestrator decisions
  • .map/<branch>/progress_<branch>.md "## Final Verification" section is for human readability only
  • Orchestrator (map-efficient.md) MUST read JSON, not parse markdown
  • Both must be written, but only JSON is used programmatically

Read the full file on GitHub · 235 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 · 235 lines · 17 tokens per session scan A 9c45122c881e

Subscribe to this mod's changes

final-verifier is an agent published in the GitHub repository azalio/map-framework (153 stars, last pushed 4d ago), licensed MIT. It adds 17 tokens to every session and 2,477 once invoked, about $0.0001 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 2026-08-30.

Related

Other agents, from other repositories

deep-paper-researcher

Token-isolated deep research agent for academic papers. Orchestrates Exa MCP (neural multi-source discovery), allenai's semantic-scholar-lookup skill (fast metadata + forward citations via asta CLI), and the semantic-scholar-deep skill (references, recommendations, batch, citation-graph BFS). Use when the user asks…

CodeAlive-AI/ai-driven-development · 210 tokens

reviewer

Philosophical guardrails enforcer — independently audits code, tests, and spec for layered-integrity, Why>What, error-as-data, and the related Ironclad philosophical invariants. Activate only when the connected project contains spec.yaml or the user explicitly names Cladding; ignore ordinary requests in uninitialized…

qwerfunch/cladding · 68 tokens

developer

Implementer — writes production code, tests, and migrations. The "generic engineer" fallback when no narrower specialist exists. Activate only when the connected project contains spec.yaml or the user explicitly names Cladding; ignore ordinary requests in uninitialized projects.

qwerfunch/cladding · 51 tokens

orchestrator

Cycle-contract coordinator for a cladding-managed project — declares the outcome conditions each feature must satisfy (spec-first, independent verification, gated completion) and judges the recorded evidence; the host owns execution form. Activate only when the connected project contains spec.yaml or the user…

qwerfunch/cladding · 69 tokens

planner

SSoT custodian — keeps spec.yaml structurally clean. Adds features, archives them, and ensures EARS pattern compliance. Activate only when the connected project contains spec.yaml or the user explicitly names Cladding; ignore ordinary requests in uninitialized projects.

qwerfunch/cladding · 53 tokens

observability

Log and metrics analyst — reads .cladding/audit.log.jsonl, perf/baseline.json, and drift reports; surfaces patterns the human can act on. Activate only when the connected project contains spec.yaml or the user explicitly names Cladding; ignore ordinary requests in uninitialized projects.

qwerfunch/cladding · 61 tokens