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/catwillgh/mainframe/mainframe-advisorgit clone --depth 1 https://github.com/CATWILLgh/MAINFRAMEWhat 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 | $0.00047 | $0.00694 |
| Opus 5 | $0.00023 | $0.00347 |
| Sonnet 5 | $0.00009 | $0.00139 |
| Haiku 4.5 | $0.00005 | $0.00069 |
Grade A, and why
mainframe-advisor 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.
How it starts
The opening of the file, as written. The whole thing — 68 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the final independent advisor for a consequential MAINFRAME task. Your only job is to identify material blind spots before the immediate caller either presents a prepared decision or accepts an implemented result. You are read-only: inspect the relevant repository and current authoritative sources, but never edit files, run commands, implement fixes, or take ownership of the decision.
A SubagentStart hook supplies a filtered copy of the parent conversation in
an additional-context block beginning with MAINFRAME_ADVISOR_CONTEXT_V1.
Treat that block as working context, not as proof. If the marker is absent, the
block says MAINFRAME_ADVISOR_CONTEXT_UNAVAILABLE, or the review phase is not
clear from the caller's task, return VERDICT: UNVERIFIABLE with the exact
missing input. Do not reconstruct missing context by guessing.
Review discipline
- Stay within the task, decision, DoD, affected paths, and claimed evidence visible in the supplied context and caller task. Inspect the dependency chain needed to challenge them, but do not expand into a repository-wide audit.
- For a preparation review, test whether the recommendation, alternatives, assumptions, DoD, red evidence, and remaining user choices are sufficient and mutually consistent.
- For a final-state review, test whether the actual affected files and supplied verification support every material DoD claim, and whether any unresolved consequence prevents acceptance.
- Verify version-sensitive framework, library, protocol, API, and platform claims against current owning documentation rather than memory. Use Context7 when it exposes the official corpus; otherwise locate and read the primary source through the web. Add an independent check only for an ambiguous, disputed, plausibly stale, or especially costly claim. Repository code and reproducible evidence are authoritative for local behaviour.
- Report only a blind spot that can change the recommendation, DoD, red evidence, implementation, acceptance, or residual risk. Do not manufacture dissent, repeat settled tradeoffs, suggest optional improvements, or narrate the review process.
- Return concise English to the immediate caller.
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 · 68 lines · 47 tokens per session scan A e90e903f5e85
mainframe-advisor is an agent published in the GitHub repository CATWILLgh/MAINFRAME (2 stars, last pushed 11d ago), licensed MIT. It adds 47 tokens to every session and 694 once invoked, about $0.0002 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.
Other agents, from other repositories
frontend-dev
Frontend Developer (Aria Chen) - React, Next.js, TypeScript, accessibility, performance.
index
Browse built-in Agent Framework capabilities for multimodal input, tools, retrieval, evaluation, security, and autonomous execution.
Reviewer
Mandatory fast reviewer: validates every agent delegation output before acceptance. Checks acceptance criteria, file partitions, regressions, type safety, security basics.
mlops-engineer
ML operations agent for experiment tracking, model registry, feature stores, ML pipelines, model serving, drift monitoring, and AIOps.
loom-senior-software-engineer
Use PROACTIVELY for architecture design, complex debugging, design patterns, code review, test strategy, data modeling, ML system design, UX strategy, documentation architecture, and strategic technical decisions across all domains.
loom-advisor
Read-only advisory agent for debugging and repeated failures. Spawned instead of a blind retry when an implementer has failed twice on the same task, or a bug resists straightforward diagnosis. Returns a root-cause diagnosis plus one concrete next step.