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/tacoda/keystone/quality-radargit clone --depth 1 https://github.com/tacoda/keystoneWrote 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/tacoda/keystone/quality-radar)<a href="https://agentmods.dev/agents/tacoda/keystone/quality-radar"><img src="https://agentmods.dev/badge/agents/tacoda/keystone/quality-radar.svg" alt="Measured on agentmods" 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 | $0.00027 | $0.00569 |
| Opus 5 | $0.00014 | $0.00284 |
| Sonnet 5 | $0.00005 | $0.00114 |
| Haiku 4.5 | $0.00003 | $0.00057 |
Grade A, and why
quality-radar 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 3d 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 — 34 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Sensor: quality-radar
Aggregates verification and review sensors into a five-dimension scorecard for the current diff (or the whole codebase, on audit).
- Trigger — audit (codebase-wide) and review (diff-scoped).
- Inputs — outputs of lint, type-check, test, coverage, review-security, review-functional, plus complexity metrics from risk-fingerprint.
- Exit condition — every dimension has a score and at least one cited finding (or an explicit "no findings").
- Output — one score per dimension (
green | yellow | red) with the citations that drove the score. - State writes — proposes a diff to
corpus/state/quality-radar.md. User accepts or edits.
Dimensions
| Dimension | Driven by | Green | Yellow | Red |
|---|---|---|---|---|
| Type safety | type-check exit + signature churn |
no errors | errors confined to one region | errors across regions or any/unknown proliferation |
| Test quality | test exit + coverage + test-to-code ratio in diff |
passes; coverage stable or up; new code tested | passes; coverage flat with new code | failures, coverage drop, or new code untested |
| Readability | lint + complexity from risk-fingerprint |
lint clean; no new critical-complexity regions | lint warns; one new high-complexity region | lint errors or multiple new critical regions |
| Security | review-security findings |
none | low-severity only | medium or higher |
| Performance | hot-path heuristics from review-functional + diff size on perf-critical regions (flagged in CODEBASE_STATE.md) |
no perf-critical touch or no concerns | perf-critical touch, no benchmarks | perf-critical touch with regression signal |
Not a gate
This is a scorecard, not a verification gate. Red on a dimension is a signal to discuss, not an automatic block — the review action decides what to do with it. The actual block is whatever the underlying sensor already imposes (a failed test blocks regardless of how the radar shades the diff).
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.
- 3d ago First seen · 34 lines · 27 tokens per session scan A e896fe8c036f
quality-radar is an agent published in the GitHub repository tacoda/keystone (44 stars, last pushed 2mo ago), licensed MIT. It adds 27 tokens to every session and 569 once invoked, about $0.0001 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 2026-08-30.
Other agents, from other repositories
c-cpp-expert-agent
Deep C++ expertise for template metaprogramming, SFINAE, constexpr, type traits, and memory model questions in embedded context.
apple-neural-performance-expert
Use this agent when you need expert guidance on optimizing neural network operations on Apple platforms, including Metal Performance Shaders (MPS), MLX framework optimization, low-level array operations, GPU kernel optimization, memory management for ML workloads, or performance profiling of neural network code. This…
firmware-reviewer
IoT/embedded specialist pre-implementation reviewer. Outputs threat model TM-{slug}.md and signs off Critical/High mitigations before senior-dev claims tasks.
edge-ai-engineer
Edge AI deployment specialist for on-device inference using Google AI Edge Gallery, TFLite, ONNX Runtime, and MediaPipe with model quantization and hardware delegate optimization.
amk-megakernel-optimizer
Use to autonomously optimize a model's AMK megakernel, runs the correctness-gated propose->eval->keep/revert loop (and the unattended autoresearch driver) and reports the measured, drift-robust speedup over AMK's own baseline.
neuron-framework-autoport-vllm-agent
Autonomous agent for porting HuggingFace models to the vLLM-Neuron Trainium2 backend. Accepts a model name and HuggingFace model ID, then executes the full porting workflow: architecture research, code generation, model registration, and validation. Context: User wants to port Yi to vLLM-Neuron user: "Port yi…