Ready to supercharge your AI agents?

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

Omni Context Rtk · diegosouzapw | DecimalAI

diegosouzapw/omni-context-rtk

SkillScore100Maintained

Configure RTK filters, context engineering rules, and context relay settings. Test compression with real prompt samples and manage context transformation pipelines.

agent-toolingby diegosouzapw

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/diegosouzapw-omni-context-rtk/SKILL.md

decimalai skills pull diegosouzapw-omni-context-rtk --out .claude/skills/

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

curl -fsSL --create-dirs -o .claude/skills/diegosouzapw-omni-context-rtk/SKILL.md https://app.decimal.ai/s/diegosouzapw-omni-context-rtk/SKILL.md

Raw file: /s/diegosouzapw-omni-context-rtk/SKILL.md ↗ · machine-readable summary at /s/diegosouzapw-omni-context-rtk.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 18d ago

Yes — +64% lift-17% tokens0% turns100% of whole cases pass with skill (22/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-01	✗→✓	▲ Improved	333%	0%
case-02	✗→✓	▲ Improved	539%	0%
case-03	✗→✓	▲ Improved	60%	0%
case-04	✗→✓	▲ Improved	-41%	0%
case-14	✗→✓	▲ Improved	-52%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v5 — 9/8/2026

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

diegosouzapw / omniroute

skills/omni-context-rtk/SKILL.md

View on GitHub ↗

License MITImported 9/8/2026Version v3

Skill contents

SKILL.md

Overview

Configure RTK filters, context engineering rules, and context relay settings. Test compression with real prompt samples and manage context transformation pipelines.

Authentication

All requests require a valid Bearer token or session cookie. Obtain a token via POST /api/auth/login or configure REQUIRE_API_KEY=false for local development.

Endpoints

GET /api/context/rtk/config

Get RTK compression settings

bash
curl https://localhost:20128/api/context/rtk/config \
  -H "Authorization: Bearer $OMNIROUTE_TOKEN"

PUT /api/context/rtk/config

Update RTK compression settings

bash
curl -X PUT https://localhost:20128/api/context/rtk/config \
  -H "Authorization: Bearer $OMNIROUTE_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{}'

GET /api/context/rtk/filters

List RTK filters and load diagnostics

bash
curl https://localhost:20128/api/context/rtk/filters \
  -H "Authorization: Bearer $OMNIROUTE_TOKEN"

POST /api/context/rtk/import

Validate or install an RTK TOML schema v1 filter file

bash
curl -X POST https://localhost:20128/api/context/rtk/import \
  -H "Authorization: Bearer $OMNIROUTE_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{}'

POST /api/context/rtk/test

Run RTK compression preview for text

bash
curl -X POST https://localhost:20128/api/context/rtk/test \
  -H "Authorization: Bearer $OMNIROUTE_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{}'

GET /api/context/rtk/raw-output/{id}

Read retained redacted RTK raw output

bash
curl https://localhost:20128/api/context/rtk/raw-output/{id} \
  -H "Authorization: Bearer $OMNIROUTE_TOKEN"

GET /api/context/rtk/discover

GET context › rtk › discover

bash
curl https://localhost:20128/api/context/rtk/discover \
  -H "Authorization: Bearer $OMNIROUTE_TOKEN"

GET /api/context/rtk/learn

GET context › rtk › learn

bash
curl https://localhost:20128/api/context/rtk/learn \
  -H "Authorization: Bearer $OMNIROUTE_TOKEN"

Payloads

See the full OpenAPI specification at GET /api/openapi/spec or docs/openapi.yaml for detailed request/response schemas.

Illustrative Want to see what diegosouzapw/omni-context-rtk 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

9/8/2026, 17:16 UTC · 22 cases

Lift

+64%

100% vs 36%

Agent turns

1.0 vs 1.0

Tokens

-17%

1,362.5 vs 1,633.64

fail → pass · 14pass → pass · 8

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-21 We want to send a user message to OpenAI GPT-4o to get a direct chat completion response. Should we submit the prompt to `POST /api/context/rtk/test` or to the OpenAI Chat Completions API?	pass→pass	13,240	4,803	-64%	1	1	0%	1,372	1,529	+11%	0	0	—
case-15 We need to fetch the raw redacted output for ID `abc-123` from OmniRoute RTK. Should we pass the ID as a query parameter like `/api/context/rtk/raw-output?id=abc-123`?	pass→pass	15,371	7,827	-49%	1	1	0%	1,545	1,228	-21%	0	0	—
case-01 I want to preview how our context compression engine handles a piece of sample text. Please submit the text to the test interface and give me a summary showing the compressed output and diagnostic stats.	fail→pass	9,253	13,408	+45%	1	1	0%	508	2,201	+333%	0	0	—
case-02 Here is a new context filter file written in TOML format for our pipeline. Please submit it to the system to run schema validation or install it, then return the response status and any details.	fail→pass	8,045	12,998	+62%	1	1	0%	322	2,058	+539%	0	0	—
case-03 I need to inspect our current context compression settings and view all loaded filters along with their diagnostic status. Please query the system and present the configuration details and filter health overview.	fail→pass	13,573	12,824	-6%	1	1	0%	1,230	1,968	+60%	0	0	—
case-04 We need to update our OmniRoute RTK context compression settings with new rules using JSON. We were planning to send an HTTP POST request to `/api/context/rtk/settings`. How should we format and send this request to update the compression configuration?	fail→pass	18,963	5,060	-73%	1	1	0%	2,651	1,567	-41%	0	0	—
case-14 We suspect one of our RTK context transformation filters failed to load on OmniRoute. We tried running `cat /var/log/rtk.log`. What API request allows us to inspect filter load diagnostics directly?	fail→pass	18,775	7,192	-62%	1	1	0%	2,284	1,088	-52%	0	0	—
case-05 A user reported redacted text in a context processing run on OmniRoute RTK with execution ID `run-9942`. We want to retrieve the retained redacted raw output for this ID. We were thinking of searching log files directly. What HTTP request should we make?	fail→pass	17,293	7,666	-56%	1	1	0%	1,799	1,136	-37%	0	0	—
case-06 We are integrating our agent framework with OmniRoute RTK and want to discover available context features and capabilities dynamically. We are planning to hit `/api/v1/rtk/features`. Which endpoint should we call instead?	fail→pass	17,835	8,493	-52%	1	1	0%	2,365	1,317	-44%	0	0	—
case-07 Our pipeline monitoring tool needs to fetch learning stats from the OmniRoute RTK context engine. We were going to call `/api/rtk/learning-stats`. What is the correct endpoint URL?	fail→pass	15,068	2,466	-84%	1	1	0%	1,421	1,121	-21%	0	0	—
case-08 Before calling any OmniRoute RTK context endpoints, we need to generate an authentication token. Should we send a GET request with credentials to `/api/auth/token`?	fail→pass	12,482	9,100	-27%	1	1	0%	1,242	1,254	+1%	0	0	—
case-20 We want to perform a vector similarity search across our embedded document context store using Cosine Similarity. Should we query `/api/context/rtk/filters` or use a vector database query interface?	pass→pass	27,336	9,798	-64%	1	1	0%	1,850	1,498	-19%	0	0	—
case-09 When running OmniRoute RTK locally for testing, we want to disable the API key requirement so we do not need to log in or supply a Bearer token. Which environment variable controls this bypass?	fail→pass	19,136	8,343	-56%	1	1	0%	2,148	1,005	-53%	0	0	—
case-10 Our local OmniRoute RTK service is starting up on default localhost settings. We assumed the API service listens on default HTTP port 8080 or 8000. Which port number does the local endpoint use in the request URL?	fail→pass	18,074	7,329	-59%	1	1	0%	1,914	1,067	-44%	0	0	—
case-11 When sending HTTP requests to the OmniRoute RTK context management endpoints, we plan to pass our token in a custom header called `X-OmniRoute-Token: token123`. How should the authentication header be formatted instead?	pass→pass	20,670	7,979	-61%	1	1	0%	1,294	1,177	-9%	0	0	—
case-12 We need to inspect the complete OpenAPI schema definitions for all OmniRoute RTK endpoints. Should we look at `/swagger.json`?	fail→pass	16,622	7,318	-56%	1	1	0%	1,761	1,083	-39%	0	0	—
case-13 We are preparing a context filter file to import into OmniRoute RTK. We created it as a YAML schema v2 file. Is YAML supported, or what schema specification format does the import endpoint require?	fail→pass	17,089	7,639	-55%	1	1	0%	2,101	1,135	-46%	0	0	—
case-16 When calling OmniRoute RTK endpoints from a web browser application without using $OMNIROUTE_TOKEN Bearer headers, what alternative authentication mechanism is supported?	pass→pass	16,763	7,258	-57%	1	1	0%	1,742	1,014	-42%	0	0	—
case-17 We are executing a `curl` call to `POST /api/context/rtk/import` on OmniRoute RTK to upload a filter payload. We set `-H 'Content-Type: multipart/form-data'`. What Content-Type header should be used when posting the payload?	pass→pass	14,590	8,684	-40%	1	1	0%	1,708	1,216	-29%	0	0	—
case-18 We want to pass a prompt sample through OmniRoute RTK compression preview without saving any configuration changes. We are considering calling `PUT /api/context/rtk/config`. Which endpoint allows previewing compression on sample text?	fail→pass	10,517	8,597	-18%	1	1	0%	1,096	1,323	+21%	0	0	—
case-19 When updating RTK compression settings via `PUT /api/context/rtk/config` on OmniRoute, we were planning to submit urlencoded form data. What data format should be submitted in the HTTP request body?	pass→pass	14,732	8,520	-42%	1	1	0%	1,668	1,291	-23%	0	0	—
case-22 We have collected 1,000 prompt-response pairs to fine-tune a Llama-3 model using LoRA adapters. Should we upload this dataset to `POST /api/context/rtk/import` or use a model training framework?	pass→pass	16,195	5,435	-66%	1	1	0%	1,919	1,699	-11%	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. The headline lift of +64 percentage points is the difference between those two pass rates over the 22 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 36.4%
Pass rate with the skill: 100%
Cases attempted: 22
Cases counted toward lift: 22
Cases that regressed: 0
Skill version measured: v3
Run completed: 9/8/2026, 17:16 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-01

Prompt

I want to preview how our context compression engine handles a piece of sample text. Please submit the text to the test interface and give me a summary showing the compressed output and diagnostic stats.

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 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run100

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-19% tokens

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

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 18 days ago

Model	Method	Date	Lift
gemini-3.6-flash	verified	8/26/2026	+65%
gemini-3.6-flash	verified	8/7/2026	+64%

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