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/plasma-ai/fractal/models_and_effortgit clone --depth 1 https://github.com/plasma-ai/fractalWrote 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/plasma-ai/fractal/models_and_effort)<a href="https://agentmods.dev/agents/plasma-ai/fractal/models_and_effort"><img src="https://agentmods.dev/badge/agents/plasma-ai/fractal/models_and_effort.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 | $0.00007 | $0.01344 |
| Opus 5 | $0.00003 | $0.00672 |
| Sonnet 5 | $0.00001 | $0.00269 |
| Haiku 4.5 | $0.00001 | $0.00134 |
Grade B, and why
features/agents/models_and_effort scanned grade B with 1 finding 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
`settings.json`, then the user's `~/.claude/settings.json`); the first file How it starts
The opening of the file, as written. The whole thing — 106 lines — stays where its author put it; the contents beside it link to each section on GitHub.
features/agents/models_and_effort
[[features/agents/_index|..]]
Override sources
A node's agent identity and its overrides are configuration: the config keys
agent (the command, extra words spliced into every invocation), provider
(the route, see [[features/agents/providers]]), model, and effort. Every
invocation builder also accepts per-call model and effort overrides, so a
single spawn can deviate from the node's configured defaults.
Effort is passed through to the provider's own effort flag unvalidated — fractal does not maintain a level vocabulary; an unknown level surfaces as the agent's own error. A hard USD budget is a separate per-invocation knob accepted only by backends that declare budget enforcement; asking a non-enforcing backend for one is an error.
Per-backend spelling
Each backend translates the same two overrides into its own CLI dialect:
| backend | model flag | effort flag |
|---|---|---|
claude |
--model |
--effort |
codex |
-m |
-c model_reasoning_effort="<lvl>" |
grok |
-m |
--reasoning-effort |
opencode |
-m |
--variant (a model variant) |
omp |
--model |
--thinking (a thinking level) |
For claude the effort flag outranks the settings-file effort level, and it is
the only effort channel fractal uses -- the CLAUDE_CODE_EFFORT_LEVEL env knob
is deliberately not set, since a spawn-env knob leaks into the session's own
subprocesses and would override any nested claude run; ambient copies of the
effort knobs are actively unset at invocation compose -- an operator shell's
CLAUDE_CODE_EFFORT_LEVEL would override the step pin inside the session, and a
stale CLAUDE_EFFORT (claude's own export to hook/Bash subprocesses) would
masquerade as the child session's -- and so is the model-forcing
CLAUDE_CODE_SUBAGENT_MODEL, which would force every fan-out sub-agent onto one
model, explicit per-agent pins included (the seeded node settings deliberately
do not set it either); node settings (permissions, model, environment) ride a
CLI flag rather than a file merge. On the openrouter route a model-less claude
invocation pins an explicit model slug, because the process environment beats
the settings file and the route must not trust latest-model aliases.
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 · 106 lines · 7 tokens per session scan B 30e02e360063
features/agents/models_and_effort is an agent published in the GitHub repository plasma-ai/fractal (712 stars, last pushed today), licensed Apache-2.0. It adds 7 tokens to every session and 1,344 once invoked, about $0.0000 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
analyzer
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
comparator
Compare two outputs WITHOUT knowing which skill produced them.
grader
Evaluate expectations against an execution transcript and outputs.
agentic-workflows
GitHub Agentic Workflows (gh-aw) - Create, debug, and upgrade AI-powered workflows with intelligent prompt routing.