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/nafjan/summon/zai-glm-5-3-flashgit clone --depth 1 https://github.com/Nafjan/summonWrote 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/nafjan/summon/zai-glm-5-3-flash)<a href="https://agentmods.dev/agents/nafjan/summon/zai-glm-5-3-flash"><img src="https://agentmods.dev/badge/agents/nafjan/summon/zai-glm-5-3-flash.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 5.1 | $0.00000 | $0.00316 |
| Opus 5 | $0.00000 | $0.00158 |
| Sonnet 5 | $0.00000 | $0.00063 |
| Haiku 4.5 | $0.00000 | $0.00032 |
Grade A, and why
zai-glm-5-3-flash 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 6d 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
Z.AI Coding Plan GLM-5.3 Flash (direct text seat)
This optional direct Z.AI Coding Plan seat makes one OpenAI-compatible text request to the coding-only endpoint. It does not provide a tool loop, filesystem access, session resume, or a sandbox. Give it self-contained text such as a design question, pasted diff, or review packet; use the OpenCode Z.AI seat for toolful work in a disposable boundary.
The model is explicitly requested, but treat it as a named-model result only
when the provider returns a matching model.served with reported evidence.
The public definition contains no key, account information, endpoint override,
or local configuration path. Explicit ZAI_CODING_API_KEY takes precedence;
an existing official helper configuration is read only through a bounded,
non-symlink scalar reader and never copied to a child process or receipt.
Availability, quota, pricing, context limits, and provider model routing are live facts. No fallback to another endpoint or model occurs automatically.
STATUS: DONE | PARTIAL | BLOCKED SUMMARY: FINDINGS: FOLLOW-UP: HANDOFF: LEFT_BEHIND:
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.
- 6d ago First seen · 34 lines · 0 tokens per session scan A 8ed84b712726
zai-glm-5-3-flash is an agent published in the GitHub repository Nafjan/summon (4 stars, last pushed 3d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 316 tokens. 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
ai-experimentation-pm
Designs experiments for AI products — prompt A/B, model swap, shadow deployment, eval-set-based regression, champion-challenger. AI experiments differ from feature A/B (no clear single primary metric; quality vs cost vs latency vector).
ai-feedback-loop-designer
Designs the user-correction → retraining loop. Specifies implicit + explicit feedback capture, signal-to-noise filtering, the path from "user fixed it" to "model gets better", and the cadence of re-training. The compound-interest engine of AI products.
ai-product-strategist
Strategy for AI-heavy products. Picks the right bets — model-vs-prompt architecture, build-vs-buy on models, data-moat assessment, commoditization risk, capability-vs-feature framing. Authors AI-product strategy docs distinct from standard product-strategist.
ai-roadmap-planner
Authors the 3-layer AI roadmap — DATA layer, MODEL layer, PRODUCT layer — with explicit dependencies between them. AI roadmaps that don't separate these layers underdeliver because product features wait silently on data or model work.
ai-rollback-strategist
Designs the fallback + rollback architecture for AI products. When the model fails (quality drop, cost spike, safety event, vendor outage), what does the user see and how does the system recover. Without this, AI products have brittle launches.
data-strategist
Authors the data strategy for AI products — acquisition, labeling, privacy boundaries, synthetic vs real, data moat assessment, training-data lifecycle. Without this, AI-product strategy is built on assumed data.