Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Torchserve Config Generator · dicklesworthstone | DecimalAI

dicklesworthstone/torchserve-config-generator

NewSkillScoreMaintained

Generate torchserve config generator operations. Auto-activating skill for ML Deployment. Triggers on: torchserve config generator, torchserve config generator Part of the ML Deployment skill category. Use when configuring systems or services. Trigger with phrases like "torchserve config generator", "torchserve generator", "torchserve".

by dicklesworthstone

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

AI Engineer pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/dicklesworthstone-torchserve-config-generator/SKILL.md

decimalai skills pull dicklesworthstone-torchserve-config-generator --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/dicklesworthstone-torchserve-config-generator/SKILL.md https://app.decimal.ai/s/dicklesworthstone-torchserve-config-generator/SKILL.md

Raw file: /s/dicklesworthstone-torchserve-config-generator/SKILL.md ↗ · machine-readable summary at /s/dicklesworthstone-torchserve-config-generator.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 23d ago

—+36% tokens0% turns73% of whole cases pass with skill (16/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-08	✗→✓	▲ Improved	63%	0%
case-19	✗→✓	▲ Improved	32%	0%
case-16	✓→✗	▼ Worse	35%	0%
case-01	✓→✓	＝ Same ✓	21%	0%
case-02	✓→✓	＝ Same ✓	2%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/10/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

dicklesworthstone / pi_agent_rust

tests/ext_conformance/artifacts/plugins-community/skills/08-ml-deployment/torchserve-config-generator/SKILL.md

View on GitHub ↗

License MITImported 8/10/2026Version v1

Skill contents

SKILL.md

Torchserve Config Generator

Overview

This skill provides automated assistance for torchserve config generator tasks within the ML Deployment domain.

When to Use

This skill activates automatically when you:

Mention "torchserve config generator" in your request
Ask about torchserve config generator patterns or best practices
Need help with machine learning deployment skills covering model serving, mlops pipelines, monitoring, and production optimization.

Instructions

Provides step-by-step guidance for torchserve config generator
Follows industry best practices and patterns
Generates production-ready code and configurations
Validates outputs against common standards

Examples

Example: Basic Usage Request: "Help me with torchserve config generator" Result: Provides step-by-step guidance and generates appropriate configurations

Prerequisites

Relevant development environment configured
Access to necessary tools and services
Basic understanding of ml deployment concepts

Output

Generated configurations and code
Best practice recommendations
Validation results

Error Handling

| Error | Cause | Solution | |-------|-------|----------| | Configuration invalid | Missing required fields | Check documentation for required parameters | | Tool not found | Dependency not installed | Install required tools per prerequisites | | Permission denied | Insufficient access | Verify credentials and permissions |

Resources

Official documentation for related tools
Best practices guides
Community examples and tutorials

Related Skills

Part of the ML Deployment skill category. Tags: mlops, serving, inference, monitoring, production

Illustrative Want to see what dicklesworthstone/torchserve-config-generator would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/29/2026, 19:07 UTC · 22 cases

Measured on 21 comparable cases — below the 22 needed to publish a lift, so this result is shown here but not on the registry card or leaderboard.

Lift

50% vs 50%

Agent turns

1.0 vs 1.0

Tokens

+36%

2,095.52 vs 1,545.9

fail → pass · 2pass → pass · 14fail → fail · 5pass → fail · 1

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 I am writing a custom PyTorch model handler in Python for TorchServe to pre-process incoming image tensors and run inference through a ResNet50 model. Generate the complete Python handler script extending BaseHandler with initialize, preprocess, and inference methods.	pass→pass	16,229	21,717	+34%	1	1	0%	3,299	3,998	+21%	0	0	—
case-02 I need to deploy an ONNX format model on Triton Inference Server. Generate the Triton config.pbtxt configuration file defining input tensors input__0 of shape [-1, 3, 224, 224] and output tensor output__0 of shape [-1, 1000].	pass→pass	11,607	8,632	-26%	1	1	0%	1,841	1,877	+2%	0	0	—
case-03 I need to package a PyTorch model checkpoint model.pt and extra file index.json into a TorchServe .mar archive file using the torch-model-archiver command line tool. Provide the complete torch-model-archiver CLI command with appropriate flags.	pass→pass	7,090	9,434	+33%	1	1	0%	1,383	1,803	+30%	0	0	—
case-04 Create a Helm chart values file values.yaml for deploying a generic model serving application on a Kubernetes cluster with autoscaling enabled based on CPU utilization exceeding 80%.	pass→pass	15,977	21,096	+32%	1	1	0%	2,923	3,810	+30%	0	0	—
case-05 We are containerizing TorchServe in a Docker container behind an external API gateway. The container needs to accept public inference requests on port 8080 from outside the container, but the management REST API on port 8081 must remain restricted to local loopback (127.0.0.1) for security. Generate the full config.properties configuration file.	pass→pass	15,324	8,354	-45%	1	1	0%	2,440	1,920	-21%	0	0	—
case-06 We are deploying TorchServe on an AWS g4dn.12xlarge instance equipped with 4 NVIDIA T4 GPUs. We want TorchServe to utilize all 4 GPUs and run 2 model worker processes per GPU by default. Generate the required config.properties file setting these GPU resource allocations.	fail→fail	21,955	12,844	-41%	1	1	0%	1,944	2,954	+52%	0	0	—
case-07 Our Kubernetes monitoring cluster scrapes metrics in Prometheus format. We need TorchServe to export metrics in Prometheus format on port 8082 instead of the standard log file format. Generate the config.properties snippet for metric collection.	fail→fail	15,256	8,136	-47%	1	1	0%	1,477	1,948	+32%	0	0	—
case-08 We have packaged two model archives (resnet50.mar and bert_ner.mar) into /opt/ml/models. We want TorchServe to start up, load its model archives from /opt/ml/models, and automatically pre-load both models on server boot. Generate the complete config.properties file.	fail→pass	18,929	10,116	-47%	1	1	0%	1,433	2,337	+63%	0	0	—
case-09 Our TorchServe instance processes high-resolution 4K medical DICOM images where single request payloads reach up to 40 MB. The default TorchServe request size limit causes HTTP 413 Payload Too Large errors. Generate a config.properties file that expands the maximum request payload size limit to 50,000,000 bytes.	pass→pass	5,406	12,422	+130%	1	1	0%	1,032	2,296	+122%	0	0	—
case-10 Our image generation model on TorchServe produces high-bitrate uncompressed audio arrays that exceed the default response size boundary of 6.5MB, causing response truncated errors. Generate the TorchServe config.properties setting that expands the maximum allowable response payload size to 100,000,000 bytes.	pass→pass	3,404	7,442	+119%	1	1	0%	631	1,438	+128%	0	0	—
case-11 A single-page web app hosted at https://app.example.com makes direct REST fetch calls to our TorchServe inference server. Browsers block these requests due to CORS policies. Generate the config.properties setting that permits cross-origin resource sharing specifically for https://app.example.com and allows GET and POST methods.	pass→pass	6,491	4,943	-24%	1	1	0%	1,141	1,339	+17%	0	0	—
case-12 Under peak load of 5,000 requests per second, synchronous disk logging in TorchServe creates an I/O bottleneck that increases inference latency p99. Generate the config.properties setting that switches TorchServe logging to asynchronous mode to eliminate disk I/O blocking.	pass→pass	5,768	6,683	+16%	1	1	0%	1,023	1,394	+36%	0	0	—
case-13 We are migrating our model infrastructure to KServe standards and require TorchServe to format REST request payloads and response outputs according to the KServe v2 protocol specification. Generate the config.properties parameter that configures this payload envelope wrapper.	fail→fail	7,077	10,156	+44%	1	1	0%	1,364	2,312	+70%	0	0	—
case-14 During sudden traffic bursts, TorchServe drops incoming inference requests when the internal queue fills up. We want to increase the maximum job queue size per model to 2,000 requests to buffer burst traffic before returning HTTP 503. Generate the config.properties parameter for this.	fail→fail	4,015	7,591	+89%	1	1	0%	789	1,647	+109%	0	0	—
case-15 Our model archives contain individual requirements.txt files with custom pip dependencies (such as specific tokenizers). We want TorchServe to automatically install these Python dependencies when loading each model archive without requiring manual pre-installation. Generate the config.properties setting for this behavior.	pass→pass	5,908	6,155	+4%	1	1	0%	975	1,426	+46%	0	0	—
case-16 To comply with internal security guidelines, all administrative access to TorchServe's management port must be encrypted with TLS/SSL using certificate file /etc/ssl/certs/torchserve.crt and private key file /etc/ssl/private/torchserve.key. Generate the config.properties file configuring TLS for the management endpoint.	pass→fail	9,013	10,241	+14%	1	1	0%	1,799	2,430	+35%	0	0	—
case-17 Our large 13B parameter LLM takes nearly 3 minutes (180 seconds) to download and load weights into GPU memory when a worker starts. TorchServe's default worker startup timeout causes worker initialization to fail and kill the process prematurely. Generate the config.properties setting that extends worker initialization timeout to 180 seconds.	fail→fail	3,927	8,006	+104%	1	1	0%	675	1,787	+165%	0	0	—
case-18 Port 8081 is already occupied by a local system daemon on our deployment host. We need TorchServe to listen for management REST API calls on port 9081 while maintaining default inference on port 8080. Generate the config.properties parameter for the management address.	pass→pass	4,919	4,587	-7%	1	1	0%	702	1,204	+72%	0	0	—
case-19 By default, TorchServe exposes system and operational metrics via a dedicated REST endpoint. Ensure the metrics API service is explicitly enabled in the config.properties file.	fail→pass	6,453	7,588	+18%	1	1	0%	1,272	1,685	+32%	0	0	—
case-20 When serving models stored in S3 or remote URIs, TorchServe can stream model artifacts directly rather than staging them through temporary directory caching. Generate the config.properties setting that enables direct download mode for remote model locations.	pass→pass	9,900	8,248	-17%	1	1	0%	1,418	1,468	+4%	0	0	—
case-21 We want to configure TorchServe model serving properties to enable batch inference with a maximum batch size of 8 requests and a maximum batch delay threshold of 100 milliseconds. Generate the model configuration properties for TorchServe.	pass→pass	8,992	12,424	+38%	1	1	0%	1,704	2,831	+66%	0	0	—
case-22 Generate a complete, production-ready config.properties file for TorchServe running on a GPU server. It must bind inference to 0.0.0.0:8080, management to 127.0.0.1:8081, metrics to 0.0.0.0:8082 with Prometheus format, specify model store at /mnt/models, use 1 GPU, enable async logging, and set max request size to 10MB (10485760 bytes).	pass→pass	12,840	9,828	-23%	1	1	0%	2,632	2,439	-7%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 21 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of 0 percentage points is the difference between those two pass rates over the 21 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 50%
Pass rate with the skill: 50%
Cases attempted: 22
Cases counted toward lift: 21
Cases that regressed: 1
Skill version measured: v1
Run completed: 8/29/2026, 19:07 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-19

Prompt

By default, TorchServe exposes system and operational metrics via a dedicated REST endpoint. Ensure the metrics API service is explicitly enabled in the config.properties file.

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signalsno signals yet

BenchmarkRun a verified benchmark with positive lift

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+26% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

No signals yet — run a benchmark or collect evaluated traces to earn a score.

Verified on gemini-3.6-flash · re-tested 22 days ago

Other measured skills in the registry, with their headline benchmark lift.