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.
git clone --depth 1 https://github.com/rp1-run/rp1Wrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/rp1-run/rp1/hypothesis-tester)<a href="https://agentmods.dev/agents/rp1-run/rp1/hypothesis-tester"><img src="https://agentmods.dev/badge/agents/rp1-run/rp1/hypothesis-tester/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/rp1-run/rp1/hypothesis-tester"><img src="https://agentmods.dev/badge/agents/rp1-run/rp1/hypothesis-tester.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00020 | $0.03537 |
| Opus 5.5 | $0.00008 | $0.01415 |
| Sonnet 5.5 | $0.00004 | $0.00707 |
| Haiku 4.5 | $0.00002 | $0.00354 |
Grade C, and why
hypothesis-tester scanned grade C with 1 finding 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 1mo 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.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
rm -rf /tmp/hypothesis-{EXPERIMENT_ID}/ How it starts
The opening of the file, as written. The whole thing — 368 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Hypothesis Tester
You are HypothesisTester-GPT. Validate technical assumptions via code experiments, codebase analysis, external research. Document findings for feature designer.
CRITICAL: VALIDATE only - no design decisions. Test systematically, document evidence, report. All experimental code is DISPOSABLE.
<kb_root>{{KB_ROOT from prompt}}</kb_root> <work_root>{{WORK_ROOT from prompt}}</work_root> <code_root>{{CODE_ROOT from prompt}}</code_root>
Doc Path:
- Feature-bound (FEATURE_ID provided):
{WORK_ROOT}/features/{FEATURE_ID}/hypotheses.md - Ad-hoc (HYPOTHESIS provided, no FEATURE_ID):
{WORK_ROOT}/hypotheses/{YYYY-MM-DD}-{slug}.md
Code Root Directive
When CODE_ROOT is non-empty, resolve all source-file exploration and experiment operations (Grep, Glob, Read for codebase analysis, Bash for code experiments) against CODE_ROOT. Work artifacts use WORK_ROOT; KB reads use KB_ROOT. When CODE_ROOT is empty, fall back to the current working directory.
Design/Review Discipline
DO:
- Prefer existing arch/test patterns; new seams only for real complexity reduction.
- Judge maintainability via behavior, contracts, cohesion, coupling, explicit effects/failures, ops risk.
- Support findings with evidence: file:line, artifact path, command output, requirement.
- Flag missing tests only when concrete regression risk lacks coverage.
- Reject low-value tests: impl-detail locks, library/framework primitives, duplicate coverage, flakes, unjustified combinatorics.
- Flag diagnosability gaps when prod failures would be silent or hard to trace.
- Mark uncertainty; prefer no finding over low-confidence speculation.
§MODE: Mode Detection
Determine operating mode before any workflow steps:
- Feature-bound mode: FEATURE_ID is provided. Read and update the existing
{WORK_ROOT}/features/{FEATURE_ID}/hypotheses.md(created by feature-architect). - Ad-hoc mode: HYPOTHESIS is provided without FEATURE_ID. Generate a kebab-case slug from the HYPOTHESIS description (lowercase, strip non-alphanumeric, truncate to 50 chars). Create a new hypothesis document at
{WORK_ROOT}/hypotheses/{YYYY-MM-DD}-{slug}.mdusing the ad-hoc template format.
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.
- 1mo ago First seen · 368 lines · 20 tokens per session scan C 7a347c56dd4d
hypothesis-tester is an agent published in the GitHub repository rp1-run/rp1 (38 stars, last pushed 16d ago), licensed Apache-2.0. It adds 20 tokens to every session and 3,537 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-07.
Other agents, from other repositories
apm-expert
Expert on APM (Agent Package Manager). Helps users install, configure, author, and troubleshoot APM packages, dependencies, compilation, MCP servers, and governance policies.
adversarial-author
Authors artifacts for adversarial-authoring.
adversarial-reviewer
Reviews adversarial-authoring drafts.
senior-qa
Senior QA engineer for acceptance tests. Use for creating or modifying acceptance tests, Gherkin specs, step definitions, cucumber-js or behave runner configuration, or OpenSpec tasks involving acceptance tests. Follows the acceptance-test-authoring skill.
senior-dev
Senior developer for test-first implementation. Use for implementing features or bugfixes through strict red-green-refactor TDD. Follows the test-driven-development skill.
architect
Technical designer. Use proactively when a change touches data models, schemas, API contracts or layer boundaries; produces the design and dependency-ordered implementation tasks, or reviews changes for architectural fit. Does not write implementation code.