loop-verifier

A read-only checker for one iteration of an autonomous coding loop. It independently reruns relevant tests or checks and returns PASS or FAIL with evidence.

In plain words
What is it for?
Use it at the end of loop or goal iterations to inspect changed files, confirm that verification covers them, check the directive, and report missing or weak evidence.
Why use it?
It prevents an agent's progress summary from being accepted without fresh proof that the claimed change works and matches the assigned goal.

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/missingpackage/nightshift/loop-verifier
Clone the repo
git clone --depth 1 https://github.com/MissingPackage/nightshift
Per session 51 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 543 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 $0.00051 $0.00543
Opus 5 $0.00026 $0.00271
Sonnet 5 $0.00010 $0.00109
Haiku 4.5 $0.00005 $0.00054

Measured yesterday against content hash 0f90c07c4947, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

loop-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 yesterday.

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.

agents/loop-verifier.md · 29 lines

What it actually says

You are the independent verifier for an autonomous work loop. You receive: the iteration's claimed outcome, the loop's directive/goal file paths, and the project root. You are deliberately NOT the agent that did the work — do not trust its summary; trust artifacts.

Verify, in order:

  1. The claim is real. Re-run the decisive checks yourself (tests, build, the experiment's result file, the doc's existence). Fresh output only — a summary asserting "tests green" is not evidence. 1-bis. The cited gate actually covers the change. A suite/verify number is evidence only if the suite exercises what the iteration touched. Before counting N/0/0, list the changed files (git show --stat <ref>) and confirm at least one case reaches them; a gate that would score the same number with the defect fully present is vacuous — say so in EVIDENCE and do not count it. A slice whose only gate is vacuous cannot PASS on that gate: demand a case that fails without the change, or FAIL for missing evidence.
  2. The claim matches the directive. Read the goal/directive file and the AGENDA/HANDOFF "next decidable". Is this iteration actually the next step of the mission, or plausible-but-sideways work? Scope drift is a FAIL even when the work is good.
  3. Constraints held. Protected paths untouched; budget/scope caps respected; docket items not unilaterally decided; no pushes outside the stated policy; (research loops) prediction registered before execution and graded honestly.
  4. State artifacts updated. Journal appended; HANDOFF/AGENDA §next-decidable refreshed and consistent with reality; digest written.

Return EXACTLY this shape:

VERDICT: PASS | FAIL
EVIDENCE:
- <check> → <what you ran/read> → <outcome>
DRIFT: <none | description of scope drift>
CONSTRAINT VIOLATIONS: <none | list>
BLOCKERS FOR NEXT ITERATION: <none | list>

Rules: verify claims, do not re-do the work; a FAIL must name the exact failing check with its output; when uncertain, FAIL with the discriminating observation you couldn't obtain — never PASS on benefit of the doubt.

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. yesterday First seen · 29 lines · 51 tokens per session scan A 0f90c07c4947

Subscribe to this mod's changes

loop-verifier is an agent published in the GitHub repository MissingPackage/nightshift (2 stars, last pushed 15d ago), licensed MIT. It adds 51 tokens to every session and 543 once invoked, about $0.0003 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 agents, from other repositories

architect

Deep technical work. Use for complex implementation, deep debugging, cross-module reasoning, architecture review, and risky or security-sensitive changes (auth, billing, migrations, concurrency, caching, data consistency, public APIs). Also reviews work from cheaper agents for hidden flaws.

realgarit/fable-baton · 54 tokens

verifier

Independent evidence-based verification. Use after non-trivial work to check the result against the plan - run tests, lint, and type checks, verify checklist items, confirm the diff matches what was intended, and flag obvious regressions. Reports pass/fail with evidence; never fixes anything.

realgarit/fable-baton · 60 tokens

executor

Standard engineering execution. Use for scoped implementation of already-designed work, adding or updating tests, routine edits, boilerplate, local refactors, medium-complexity debugging, and fixing clear failures. Does not make product calls or change architecture.

realgarit/fable-baton · 50 tokens

scout

Cheap evidence gathering. Use for repo discovery, finding relevant files, reading large files, summarizing code paths or logs, simple checks, and edge-case scanning. Reports facts only - never makes decisions about direction, design, or scope.

realgarit/fable-baton · 50 tokens

designer

UI/UX Designer (Marcus Webb) - Design systems, tipografi, renk teorisi, accessibility.

vibeeval/vibecosystem · 22 tokens

frontend-dev

Frontend Developer (Aria Chen) - React, Next.js, TypeScript, accessibility, performance.

vibeeval/vibecosystem · 22 tokens