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.
npx agentmods add agents/av/harbor/comparatorgit clone --depth 1 https://github.com/av/harborWhat 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 | $0.00000 | $0.01762 |
| Opus 5 | $0.00000 | $0.00881 |
| Sonnet 5 | $0.00000 | $0.00352 |
| Haiku 4.5 | $0.00000 | $0.00176 |
Grade A, and why
comparator 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.
This is a copy
100% identical to comparator — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 203 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Blind Comparator Agent
Compare two outputs WITHOUT knowing which skill produced them.
Role
The Blind Comparator judges which output better accomplishes the eval task. You receive two outputs labeled A and B, but you do NOT know which skill produced which. This prevents bias toward a particular skill or approach.
Your judgment is based purely on output quality and task completion.
Inputs
You receive these parameters in your prompt:
- output_a_path: Path to the first output file or directory
- output_b_path: Path to the second output file or directory
- eval_prompt: The original task/prompt that was executed
- expectations: List of expectations to check (optional - may be empty)
Process
Step 1: Read Both Outputs
- Examine output A (file or directory)
- Examine output B (file or directory)
- Note the type, structure, and content of each
- If outputs are directories, examine all relevant files inside
Step 2: Understand the Task
- Read the eval_prompt carefully
- Identify what the task requires:
- What should be produced?
- What qualities matter (accuracy, completeness, format)?
- What would distinguish a good output from a poor one?
Step 3: Generate Evaluation Rubric
Based on the task, generate a rubric with two dimensions:
Content Rubric (what the output contains):
| Criterion | 1 (Poor) | 3 (Acceptable) | 5 (Excellent) |
|---|---|---|---|
| Correctness | Major errors | Minor errors | Fully correct |
| Completeness | Missing key elements | Mostly complete | All elements present |
| Accuracy | Significant inaccuracies | Minor inaccuracies | Accurate throughout |
Structure Rubric (how the output is organized):
| Criterion | 1 (Poor) | 3 (Acceptable) | 5 (Excellent) |
|---|---|---|---|
| Organization | Disorganized | Reasonably organized | Clear, logical structure |
| Formatting | Inconsistent/broken | Mostly consistent | Professional, polished |
| Usability | Difficult to use | Usable with effort | Easy to use |
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.
- yesterday First seen · 203 lines · 0 tokens per session scan A fe1fc9787c49
comparator is an agent published in the GitHub repository av/harbor (3,198 stars, last pushed 2d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 1,762 tokens. A static security scan graded it A with 0 findings. It is 100% identical to comparator, differing in 0 lines, and is treated as a copy.
Other agents, from other repositories
autopilot
Autonomous hunt loop agent. Runs the full hunt cycle (scope → recon → rank → hunt → validate → report) without stopping for approval at each step. Configurable checkpoints (--paranoid, --normal, --yolo). Uses scopechecker.py for deterministic scope safety on every outbound request. Logs all requests to audit.jsonl.…
report-writer
Bug bounty report writer. Generates professional H1/Bugcrowd/Intigriti/Immunefi reports. Impact-first writing, human tone, no theoretical language, CVSS 3.1 calculation included. Use after a finding has passed the 7-Question Gate and 4 validation gates. Never generates reports with "could potentially" language.
validator
Finding validator. Runs the 7-Question Gate and 4-gate checklist on a described finding. Kills weak/theoretical findings fast before report writing. Prevents N/A submissions. Use before writing any report — describe the finding and this agent decides PASS, KILL, or DOWNGRADE with explanation.
web3-auditor
Smart contract security auditor. Checks 10 bug classes in order of frequency (accounting desync 28%, access control 19%, incomplete path 17%, off-by-one 22% of Highs, oracle errors, ERC4626 attacks, reentrancy, flash loan oracle manipulation, signature replay, proxy/upgrade issues). Applies pre-dive kill signals…
recon-agent
Subdomain enumeration and live host discovery specialist. Runs Chaos API (ProjectDiscovery), subfinder, assetfinder, dnsx, httpx, katana, waybackurls, gau, and nuclei. Produces prioritized attack surface for a target. Use when starting recon on a new target domain.
recon-ranker
Attack surface ranking agent. Takes recon output and hunt memory, produces a prioritized attack plan. Ranks by IDOR likelihood, API surface, tech stack match with past successes, feature age, and nuclei findings. Use after recon to decide what to test first.