TRACER is a routing system that uses traditional machine-learning models to handle predictable LLM classification requests and sends uncertain cases back to the LLM. It is for reducing the number of LLM calls while checking that the smaller model remains aligned with the teacher model and improving the routing policy from new traces. The catalogue instruction covers how to use and operate this routing workflow.
Borrowing it
Nothing to install: this file belongs to adrida/tracer. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/adrida/tracer/main/AGENTS.mdgit clone --depth 1 https://github.com/adrida/tracerWrote 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/instructions/adrida/tracer/agents-md)<a href="https://agentmods.dev/instructions/adrida/tracer/agents-md"><img src="https://agentmods.dev/badge/instructions/adrida/tracer/agents-md/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/instructions/adrida/tracer/agents-md"><img src="https://agentmods.dev/badge/instructions/adrida/tracer/agents-md.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 | $0.01915 | $0.01915 |
| Opus 5.5 | $0.00766 | $0.00766 |
| Sonnet 5.5 | $0.00383 | $0.00383 |
| Haiku 4.5 | $0.00192 | $0.00192 |
Grade A, and why
tracer AGENTS.md 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 1mo 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 — 198 lines — stays where its author put it; the contents beside it link to each section on GitHub.
TRACER integration notes
TRACER (Trace-Based Adaptive Cost-Efficient Routing) turns fixed-label
teacher decisions from traces into task-specific ML students. In this hybrid
architecture, students learn easy, repetitive decisions; calibrated deferral
retains the teacher for requests the students decline. The teacher can be an
LLM, a foundation model for System 1 decisions, or another classifier. It
supplies the reference labels in the traces. A fitted student predicts labels
directly from embeddings; an acceptance policy and optional distance guard
control which predictions to return. The caller handles deferred requests,
optionally through its teacher. Student fitting and inference can run on CPU;
fit() trains the classifier and policy without fine-tuning the teacher or
semantic encoder. Read the overview,
API reference and change log for the current scope.
Source and release status
These instructions describe this repository checkout. Final-policy certification,
staged artifact publication and strict OOD-loading corrections are unreleased
source changes. Do not imply they are already present in the PyPI package merely
because pip install tracer-llm succeeds. To use the corrected source, install
from a checkout containing the changes:
python -m pip install -e .
Python 3.12+ is required. Core dependencies are NumPy, scikit-learn and joblib. The library is MIT licensed: local fitting, saved artifacts and self-hosted prediction do not require a Tracer account or license fee. Embeddings, teacher calls and hosting still have their own resource costs. This checkout supplies a Python API and a small HTTP server; the JavaScript package records traces. Do not invent a command-line interface, a general chat model, an OpenRouter listing or parity with a separately deployed hosted service.
Identify the decision contract first
Establish the label vocabulary, input representation, teacher and quality target from the user's existing task. Preserve their selected quality requirement. Training examples have this shape:
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.
- 1mo ago Changed · -10 lines · +60 tokens per session f95d5170922e
- 1mo ago Changed · -1 lines · +13 tokens per session ee012ce13184
- 2mo ago First seen · 209 lines · 1,842 tokens per session scan A cdb9eea6caea
tracer AGENTS.md is an instructions file published in the GitHub repository adrida/tracer (1,036 stars, last pushed 1mo ago), licensed MIT. It adds 1,915 tokens to every session, about $0.0077 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.
tracer AGENTS.md compared with instructions from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| tracer AGENTS.mdadrida/tracer | Instructions file | 1.0k | 1mo ago | MIT | A | 1,915 | Codex, OpenCode |
| vscode buildNext.instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 9,719 | GitHub Copilot |
| next.js AGENTS.mdvercel/next.js | Instructions file | 143k | today | MIT | A | 7,461 | Codex, OpenCode |
| langchain AGENTS.mdlangchain-ai/langchain | Instructions file | 149k | yesterday | MIT | A | 4,680 | Codex, OpenCode |
| vscode oss-third-party-notices.instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 5,516 | GitHub Copilot |
| vscode copilot-instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 3,093 | GitHub Copilot |
| deepseek-harness AGENTS.mddeepseek-ai/deepseek-harness | Instructions file | 274k | 2d ago | MIT | A | 4,313 | Codex, OpenCode |