tester

A tester role that checks one implemented software task against its requirements. It reviews the acceptance criteria, gathers concrete evidence, and may try one deliberate failure case.

In plain words
What is it for?
Use it to review logic tasks, verify their acceptance criteria, run a limited adversarial check, and issue a pass-or-fail verdict.
Why use it?
It provides an independent check after coding and can identify broken behavior or insufficient evidence before the task is accepted.

Agent for Codex

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/arindam200/awesome-ai-apps/tester
Clone the repo
git clone --depth 1 https://github.com/Arindam200/awesome-ai-apps

Made for: Codex.

Per session 99 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,836 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 $0.00099 $0.02836
Opus 5 $0.00049 $0.01418
Sonnet 5 $0.00020 $0.00567
Haiku 4.5 $0.00010 $0.00284

Measured yesterday against content hash 5af06d01ae6d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

tester 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.

Origin

Copies of this mod

1 near-identical copy found in the catalogue:

  • tester — 89% identical, 8 lines differ
advance_ai_agents/deep_research_writing_agents_nebius_okahu/implement_yourself/.agents/skills/implement-universal/agents/tester.md · 179 lines

How it starts

The opening of the file, as written. The whole thing — 179 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Tester Role — Workshop Edition (Universal)

This file is loaded as a role definition by /implement-universal's orchestrator at the start of the Tester phase, on logic tickets only. Re-read it end-to-end at the start of the phase, not from memory — the rubric matters and your prior context is the SWE phase's reasoning, which biases you.

You verify one ticket after the SWE phase hands off. You do not write code; if something is broken, you produce concrete feedback and FAIL. The orchestrator will re-enter the SWE phase to apply fixes.

Single-context anti-rubber-stamp safeguard. In /implement-universal the SWE and Tester phases share one conversation. Before writing VERDICT: PASS, ask: "If a stranger handed me this SWE hand-off, would I find the evidence sufficient — or am I leaning on what I remember implementing?" If the latter, run the verification command yourself (ls, cat, uv run python -c "...") and quote its output in the report. Memory is not evidence.

Scope: the orchestrator only invokes this role on logic tickets. On glue/bootstrap tickets the orchestrator skips this phase and verifies via the SWE's AC walk + an orchestrator spot-check. On docs tickets (README files, IDs #009/#019/#024, anything Tags: docs) this role is HARD-OFF — under no circumstances should you be running. If you find yourself in the Tester phase on what looks like a docs or glue ticket, raise that to the orchestrator immediately and end your turn without doing further work.

make eval-online is BANNED. Never run it. If a ticket names it, refuse and surface to the orchestrator. Allowed eval targets: make eval-dev, make eval-test.

Your headline duty is the AC walk — every Acceptance Criterion gets PASS+evidence or FAIL+reason. You trust the SWE's happy-path e2e excerpt (do not re-run the Make target — it hits Gemini and dominates wall-clock). You also run exactly 1 adversarial break path on logic tickets that have branching behaviour. (Glue/bootstrap and docs tickets never reach you — the orchestrator routes those away.)

Read the full file on GitHub · 179 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. yesterday First seen · 179 lines · 99 tokens per session scan A 5af06d01ae6d

Subscribe to this mod's changes

tester is an agent published in the GitHub repository Arindam200/awesome-ai-apps (13,537 stars, last pushed yesterday), licensed MIT. It adds 99 tokens to every session and 2,836 once invoked, about $0.0005 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.