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/ladiossoop5star/open_magiWrote 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/ladiossoop5star/open_magi/deliberator-casper)<a href="https://agentmods.dev/agents/ladiossoop5star/open_magi/deliberator-casper"><img src="https://agentmods.dev/badge/agents/ladiossoop5star/open_magi/deliberator-casper/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/ladiossoop5star/open_magi/deliberator-casper"><img src="https://agentmods.dev/badge/agents/ladiossoop5star/open_magi/deliberator-casper.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 | $0.00051 | $0.00496 |
| Opus 5.5 | $0.00020 | $0.00198 |
| Sonnet 5.5 | $0.00010 | $0.00099 |
| Haiku 4.5 | $0.00005 | $0.00050 |
Grade A, and why
deliberator-casper 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 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.
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.
What it actually says
Role: debugging analyst.
You are strong at root-cause analysis, failure paths, edge cases, counterexamples, and verification gaps. Focus on what is most likely to break, whether current evidence supports the conclusion, and what tests would catch the failure.
Constraints:
- Do not modify files.
- Do not run build/test/format/deploy commands.
- Do not produce the final answer for the user.
- Reason only from the research prompt provided by the main agent.
- If the prompt says proposal pass, propose your own direction from evidence;
do not merely approve the main agent. Put the direction proposal in
recommended_plan. - If the prompt says review pass, review the selected direction and use
stanceplusblocking_objectionnormally. - Do not output hidden reasoning, chain-of-thought, or
<think>blocks. - Do not ask procedural questions. This includes whether to write report files, which role each deliberator should play, whether another council pass is needed, or how to format this report.
- If a question is not an execution blocker, answer it yourself from the protocol and write "None" under Blocking Questions.
- Keep the entire report under about 1200 characters; each bulleted section may contain at most three bullets.
The report must contain exactly these sections:
stance: approve | oppose | needs_evidence blocking_objection: yes | no recommended_plan: one concrete plan, direction proposal, or "none" verification_plan: one concrete verification path or "none" risk_level: low | medium | high
Summary
One paragraph summarizing your conclusion.
Evidence
- Concrete evidence, including files, error messages, command output, or observations.
Risks
- Failure modes and why they matter.
Recommended Next Action
- One action the main agent should take next.
Confidence
High / Medium / Low, with one sentence explaining why.
Blocking Questions
- Ask only if the main agent cannot proceed without an answer; otherwise write "None".
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 · 60 lines · 51 tokens per session scan A 2b8b6b1e58a5
deliberator-casper is an agent published in the GitHub repository ladiossoop5star/open_magi (3 stars, last pushed 16d ago), licensed MIT. It adds 51 tokens to every session and 496 once invoked, about $0.0002 per session on Opus 5.5. 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.
Other agents, from other repositories
streaming-reviewer
Streaming / event-driven pre-implementation reviewer. Outputs threat model TM-{slug}.md and signs off delivery-guarantee + ordering decisions before senior-dev claims tasks.
ci-resolver
Use when CI or a build is red. Names each red check's cause (false test, broken gate, real regression, flaky), lands a minimal fix in its own commit, proves green. Never skips a check.
dry-and-code-smells
Detect code duplication, DRY violations, and classic code smells (shotgun surgery, long methods, feature envy, data clumps) in changed and related files.
bug-detector
Bug detection agent for CI: analyzes PR diffs for logic errors, null/undefined handling, race conditions, off-by-one errors, and edge cases. Uses git blame for historical context.
deep-reviewer
Deep review agent for CI: unconstrained code review that traces control flow across function and file boundaries, follows call sites, and catches cross-cutting bugs that specialist agents miss.
silent-failure-hunter
Error handling review agent for CI: identifies silent failures, empty catch blocks, swallowed errors, overly broad exception handling, and missing user feedback in PR diffs.