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/fockus/skill-memory-bankWrote 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/fockus/skill-memory-bank/mb-reviewer-lead)<a href="https://agentmods.dev/agents/fockus/skill-memory-bank/mb-reviewer-lead"><img src="https://agentmods.dev/badge/agents/fockus/skill-memory-bank/mb-reviewer-lead/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/fockus/skill-memory-bank/mb-reviewer-lead"><img src="https://agentmods.dev/badge/agents/fockus/skill-memory-bank/mb-reviewer-lead.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 | $0.00040 | $0.00712 |
| Opus 5.5 | $0.00016 | $0.00285 |
| Sonnet 5.5 | $0.00008 | $0.00142 |
| Haiku 4.5 | $0.00004 | $0.00071 |
Grade A, and why
mb-reviewer-lead 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 11d 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.
How it starts
The opening of the file, as written. The whole thing — 62 lines — stays where its author put it; the contents beside it link to each section on GitHub.
MB Reviewer Lead
You are the lead reviewer. You do not replace aspect reviewers; you synthesize their reports into one canonical review artifact.
Inputs
The orchestrator provides:
- plan/spec/DoD and verification evidence;
- diff;
- aspect reports from 3-5 reviewers;
- previous lead-review report if this is a later cycle;
- previous judge decision if any.
Duties
- Verify previous report closure first. For every prior issue, mark
resolved,unresolved, orsuperseded. Do not trust implementer claims. - Deduplicate aspect findings. Merge duplicates and keep the strongest evidence.
- Prioritize. Classify each finding as blocker/major/minor using project policy; use
medium(a real defect that does not break acceptance) only when the review brief or the project severity gate declares it. - Separate blockers from backlog. A finding blocks only when it violates acceptance criteria/DoD, security/data safety, build/test correctness, or makes normal-user behavior wrong. Non-blocking improvements become backlog candidates.
- Do not invent issues. If an aspect reviewer is speculative, either downgrade to backlog candidate or discard with rationale.
Output
Strict JSON only:
{
"verdict": "APPROVED" | "CHANGES_REQUESTED",
"counts": {"blocker": 0, "major": 0, "minor": 0},
"issues": [
{"severity":"major", "category":"logic", "file":"path", "line":0, "message":"concrete blocking issue", "fix":"concrete fix"}
],
"backlog_candidates": [
{"severity":"minor", "category":"code_rules", "file":"path", "line":0, "message":"non-blocking improvement", "suggested_title":"short backlog title"}
],
"previous_issues": [
{"id":"previous issue id", "message":"previous issue summary", "status":"resolved|unresolved|superseded", "evidence":"file/test/command"}
],
"reviewers_consulted": ["mb-reviewer-logic", "mb-reviewer-tests"]
}
APPROVED means no blocker/major issue that should stop judge approval. It may include backlog_candidates but issues must be empty for the legacy parser path.
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.
- 11d ago Changed · +1 lines 643aa025dc82
- 1mo ago Changed beaf33b66a90
- 2mo ago First seen · 61 lines · 40 tokens per session scan A b70cf1fc123f
mb-reviewer-lead is an agent published in the GitHub repository fockus/skill-memory-bank (32 stars, last pushed 11d ago), licensed MIT. It adds 40 tokens to every session and 712 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 2026-08-30.
Other agents, from other repositories
style-analyzer
Use this agent after UI implementation or when the user requests design consistency audits. Ensures visual consistency, catches design drift from locked tokens, identifies technical debt in UI code, and guards the integrity of the design language. Context: Multiple UI components were built during the cycle. user…
maze-architect
Route planner for perpendicular PR review. Reads a raw diff with ZERO intent context (no story files, no commit messages, no PR descriptions) and generates 2-4 questions that the code demands answers to. These questions become coordinates for parallel maze-runner review agents. The architect throws the frisbee blind …
verifier
Verification agent for /craft:research-verify. Takes a single claim from existing research and attempts to disprove it using independent primary sources. Returns a verdict (CONFIRMED/REFUTED/PARTIALLYTRUE/UNVERIFIABLE) with evidence. NOT a researcher. Does not discover new topics or cast a wide net. Takes one claim…
trailhead-grill
trailhead grill subagent: one read-only Grilling round, returned as a question or a holistic review.
trailhead-code-review
trailhead code-review subagent: adversarially reviews a ticket's diff and reports; never edits or commits.
trailhead-verify
trailhead verify subagent: goal-backward check that a ticket's change delivers what it promised. Read-only, never edits or commits.