Author testSource available
browser-use
Let Jev choose browser actions
- Jev decides
- Action type, target element, and completion
- Code executes
- The browser executes actions; Mercury enters text
Inputs, sources & limits
Input · Task goal and DOM controls
The author reports a 7.092-second median after optimization across three paired runs of one flight task. Initial navigation is excluded; this is not a general browser benchmark.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
lahfir
Choose controls and actions in desktop apps
- Jev decides
- Target, action, risk, and stopping conditions
- Code executes
- Local desktop tools execute and observe again
Inputs, sources & limits
Input · A native accessibility tree and an interaction goal
Jev support lives in separate agent-desktop skill/scripts. Capturing snapshots also costs time; inference latency is not total task latency.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
droidrun
Operate interfaces on an Android phone
- Jev decides
- App, target control, and next action
- Code executes
- Mobilerun executes taps and text entry; a web view shows the trace
Inputs, sources & limits
Input · Phone state, actionable controls, and a task
The author’s Uber video reaches payment selection, not a completed booking. Dark-mode behavior has a separate demo. No real-device test was run here.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
kitze
Classify page elements and create hiding rules
- Jev decides
- Ads, promotions, popups, or content to keep
- Code executes
- An extension saves and applies visual hiding rules
Inputs, sources & limits
Input · Candidate page elements and their context
Hiding a consent dialog is neither consent nor refusal. Old rules remain on invalid model output. Browser results were not independently tested.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
moritzkremb
Choose browser actions while the user speaks
- Jev decides
- Intent, target, command completeness, and risk
- Code executes
- Local policy chooses execute, wait, confirm, or ignore
Inputs, sources & limits
Input · Live partial transcripts and page-element snapshots
Other components transcribe speech and control the browser. Author latency/cost figures were not reproduced here; spoken confirmation is not a safety guarantee.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
0xNatoshi
Choose a model for each Codex turn
- Jev decides
- Task tier and reasoning requirements
- Code executes
- Local policy selects the model, effort, and speed tier
Inputs, sources & limits
Input · The current task and earlier conversation
The roughly 60% saving is a list-price recalculation over 237 historical turns, holding token usage fixed and excluding cross-model cache loss. It is not a measured bill reduction.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
hamakyo
Connect probabilities to explicit app policies
- Jev decides
- Probabilities and scores for policy decisions
- Code executes
- Code selects automatic handling, fallbacks, or human review
Inputs, sources & limits
Input · Business state, questions, and policy configuration
A workflow-pattern library: the host executes side effects. Example thresholds need validation for each task.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
minghanminghan
Route customer messages in one decision round
- Jev decides
- Categories at each level, human handoff, and frustration
- Code executes
- Code follows the tree or hands off early
Inputs, sources & limits
Input · A customer message and predefined routing tree
Querying the whole tree in parallel is an implementation strategy. The example does not establish equivalent results across all support scenarios.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
GhalebDweikat
Filter tool output while keeping it recoverable
- Jev decides
- Whether content is still relevant to the task
- Code executes
- Keep relevant or uncertain content; cache hidden originals
Inputs, sources & limits
Input · Read, Bash, and Grep output blocks
Original output is retained when errors are detected. Other evaluation backends are supported. Results depend on the task and thresholds; not independently tested here.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
tamaratran
Trim old tool records without rewriting the rest
- Jev decides
- Which historical tool records remain useful
- Code executes
- Drop or truncate stale records and preserve the remaining text
Inputs, sources & limits
Input · Conversation, tool calls, and result references
Avoiding rewrites does not make filtering lossless. Important context may still be removed; inspect results in real conversations.
Sources checked 2026-09-19 · Not independently run here
Author testSource available
ShivamPansuriya
Load only the skill instructions a task needs
- Jev decides
- Each skill’s relevance to the current task
- Code executes
- Write skill overrides to reduce unrelated instructions
Inputs, sources & limits
Input · Project context and installed skills
The repository includes author evaluations. Saving context must not hide essential skills. Not independently tested here.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
reachjalil
Keep logs while selecting signals for deeper analysis
- Jev decides
- Diagnostic value, priority, and routing
- Code executes
- Archive logs and select content for further analysis
Inputs, sources & limits
Input · OpenTelemetry log records
The project is in preview. Reducing large-model analysis does not reduce raw archives. Business impact has not been tested here.
Sources checked 2026-09-19 · Not independently run here
Public demoPublic post only
Custom memory-system experiment
@moritzkremb
Explore lower-context memory retrieval
- Jev decides
- Memory-retrieval judgments; questions and rules are undisclosed
- Code executes
- The existing memory system uses the answers; implementation is undisclosed
Inputs, sources & limits
Input · A custom memory system; exact inputs are undisclosed
The author claims 94% fewer tokens and 2–3× faster retrieval, while calling it a quick test. Public implementation and comparable test conditions are missing.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
devagrawal09
Follow risk signals through code review
- Jev decides
- Risk, related files, evidence, and severity
- Code executes
- Structured signals organize a local review panel
Inputs, sources & limits
Input · A Git diff or an entire repository
The author describes it as experimental. Findings are review leads, not proof of defects; compiler and static-analysis integration is not yet included.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
devagrawal09
Add triage and review to coding agents
- Jev decides
- Which workflow to use and which evidence matters
- Code executes
- Tools gather evidence and return structured reports
Inputs, sources & limits
Input · Tasks, code changes, test logs, or review comments
It supplies judgments rather than taking over coding. The main agent must check reports. Test suites were not run here.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
Shifty-Eye-Games
Supervise the progress of coding workers
- Jev decides
- Whether work is stuck, complete, or needs supervision
- Code executes
- Deterministic policies schedule work; acceptance commands verify it
Inputs, sources & limits
Input · Codex work state and verification evidence
Programmer-defined acceptance commands must pass. Optimistic scores or a successful process exit are not substitutes for verification.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
ellipsis-dev
Find files through semantic tree navigation
- Jev decides
- The relevance of paths at the next level
- Code executes
- Assign multiple walkers to candidate paths
Inputs, sources & limits
Input · A natural-language question and file or directory names
It traverses the file tree with walkers, without a vector index. General retrieval accuracy or speedup has not been established here.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
jexp
Choose knowledge-graph relations step by step
- Jev decides
- The next edge and whether the goal has been reached
- Code executes
- Local beam search retains and visualizes candidate paths
Inputs, sources & limits
Input · Current node, adjacent relations, and goal
Real inference requires a valid TypeSafe key. Examples without a key use labeled fallback results. Large-graph performance has not been verified.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
realZachi
Filter database rows by meaning in SQL
- Jev decides
- Condition probabilities, categories, and ordinal scores
- Code executes
- Postgres filters, sorts, and groups the results
Inputs, sources & limits
Input · Database rows and a natural-language condition
This extension calls an external model. No index does not mean no query cost. Scale and throughput have not been tested here.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
superagents-lab
Choose search sources, then rank results
- Jev decides
- Source choice, time range, and relevance
- Code executes
- Search services fetch data; the app merges and ranks links
Inputs, sources & limits
Input · A query and candidates from multiple engines
Search1API and search engines perform retrieval. Jev selection does not guarantee complete coverage or correct results.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
reachjalil
Break large taxonomies into layered decisions
- Jev decides
- Which branch to enter at the current level
- Code executes
- Code traverses recursively to a leaf
Inputs, sources & limits
Input · A JSON taxonomy tree and user goal
Layered calls address candidate-set size, but early mistakes can affect the final category. Accuracy has not been tested here.
Sources checked 2026-09-19 · Not independently run here
Public demoPublic post only
Instant spreadsheet checks
@ctnicholasdev
Trigger multiple checks when a cell changes
- Jev decides
- The author reports more than 20 checks per edit
- Code executes
- The app shows warnings after cell edits
Inputs, sources & limits
Input · The cell being edited; specific questions are undisclosed
Only the author’s demo post was verified. Check lists, false-positive rates, and full source are unavailable; this is not a validated data-checking product.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
jkudish
Check claims, screen content, and rank candidates
- Jev decides
- Evidence support, injection risk, and relevance
- Code executes
- The caller decides whether to allow, review, or block
Inputs, sources & limits
Input · Claims and evidence, web text, or candidate lists
The README includes invocation examples. The screen tool only recommends a decision; enforcement belongs to the caller. Not independently tested here.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
andrelandgraf
Screen content before forwarding requests
- Jev decides
- Whether content contains injection or unsafe material
- Code executes
- Forward allowed requests and block other outcomes
Inputs, sources & limits
Input · HTTP request bodies
The current project is a content gate, not a task-difficulty model router. Protection coverage has not been independently tested.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
y0usaf
Observe risks before and after tool execution
- Jev decides
- Destructiveness, scope violations, leakage, and impact
- Code executes
- The extension warns or gates actions according to its mode
Inputs, sources & limits
Input · Commands, file changes, and tool output
The default shadow mode only observes and warns. Integration does not mean every dangerous action is actively blocked.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
jomatsu
Validate meaning beyond field formats
- Jev decides
- Description alignment, sensitive information, and more
- Code executes
- Thresholds convert answers into Zod issues
Inputs, sources & limits
Input · Data that has passed basic format checks
Probabilistic judgments can be wrong. Calibrate on business data; they do not replace deterministic format validation.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
NicolasMontone
Choose permission paths for AI SDK tool calls
- Jev decides
- Automatic approval, human confirmation, blocking, or relevance
- Code executes
- Apply deterministic rules first, then enforce decision-based gates
Inputs, sources & limits
Input · Task, tool definitions, and pending calls
Model failures fall back to ask-human. Applications still need to define and validate their permission policies.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
emrickgarrett
Play against Jev in a browser
- Jev decides
- Movement, view, aiming, firing, and jumping
- Code executes
- The server advances an FPS simulation and executes actions
Inputs, sources & limits
Input · Structured game state
Roughly 9 Hz is the README-described decision tick, not independently measured throughput. A deterministic fallback handles API unavailability.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
standardagents
Choose the next route for a simulated car
- Jev decides
- A feasible path and speed combination
- Code executes
- A local simulator computes geometry and vehicle control
Inputs, sources & limits
Input · Roads, traffic state, and precomputed trajectories
A browser driving simulation, not evidence of real autonomous driving. The hosted demo requires sign-in.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
leftspace89
Choose flight paths for Flappy Bird
- Jev decides
- A candidate path through the gap
- Code executes
- The game schedules and executes flap actions
Inputs, sources & limits
Input · Pipe positions and code-simulated trajectories
Usually one decision per pipe. Jev does not read the screen; this is not frame-by-frame visual control.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
phureewat29
Drive scene changes in a text adventure
- Jev decides
- Location, pacing, mood, danger, and immersion
- Code executes
- Switch music, backgrounds, and the next prompt
Inputs, sources & limits
Input · Scenes generated by a narrative model
Other models generate the story; Jev evaluates state. This is a role-playing prototype.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
lbotinelly
Coordinate takeoffs and landings at toy airports
- Jev decides
- Diversion, holding, yielding, and landing order
- Code executes
- A simulator advances aircraft and airport state
Inputs, sources & limits
Input · Fuel, routes, nearby traffic, and airport state
A toy aviation simulation. Author-reported latency was not independently measured; it says nothing about real air-traffic safety.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
cocktailpeanut
Compose multi-track music through repeated choices
- Jev decides
- Instruments, harmony, rhythm, and whole-bar patterns
- Code executes
- Code renders choices as notes, with editing and MIDI export
Inputs, sources & limits
Input · A musical description, recent bars, and a harmony plan
Not an audio-generation model. The project distinguishes real decisions, fixtures, and manual edits. Audio quality was not independently reviewed.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
AmoghCreator
Choose tactical actions in Doom
- Jev decides
- Constrained movement, aiming, and shooting actions
- Code executes
- An asynchronous controller maintains actions between model calls
Inputs, sources & limits
Input · Game variables, ray checks, and structured combat state
Engine frame rate differs from model request frequency. The author’s roughly 10 Hz figure has not been independently reproduced.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
joevidev
Assemble interfaces from a component library
- Jev decides
- Component types, fields, and styles
- Code executes
- React renders implemented components and page blocks
Inputs, sources & limits
Input · A user description and finite component/configuration choices
Jev does not freely write code. Capabilities depend on the existing component choices, not arbitrary website generation.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
adammichaelwood
Explore harmony knowledge and note selection
- Jev decides
- Voices, pitch, duration, or chord composition
- Code executes
- Rule-based scorers evaluate; code arranges and plays notes
Inputs, sources & limits
Input · SATB exercises or continuous piano context
The author says AI did much of the programming and execution. An engaging piano demo and a scientific evaluation should be judged separately.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
fhshaik
Select Mario controls from emulator state
- Jev decides
- Controller actions, jump value, and danger
- Code executes
- The emulator executes selected buttons and advances frames
Inputs, sources & limits
Input · RAM, telemetry, terrain, enemies, and latency information
The model reads structured JSON, not screenshots. The repository contains no ROM. Gameplay footage alone does not prove reliable completion.
Sources checked 2026-09-19 · Not independently run here
Author testSource available
RomanSlack
Supply tactical decisions in a drone simulator
- Jev decides
- Detours, climbing, braking, and target loss
- Code executes
- A high-frequency controller and safety layer execute or veto suggestions
Inputs, sources & limits
Input · Obstacle and target states from classical vision
Simulated in MuJoCo, not a real drone flight. Jev is neither the vision nor high-frequency control layer. Key author demos are listed as single runs.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
irfndi
Add advisory signals to liquidity rules
- Jev decides
- Distribution, toxic flow, recovery, and market stress
- Code executes
- Record judgments and compare them with deterministic rules
Inputs, sources & limits
Input · Pool state, trading flow, and market signals
The source explicitly states shadow/advisory only: Jev does not drive ENTER/EXIT. This is not verified Jev-controlled trading or profitability.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
AboveColin
Turn home state into usable judgments
- Jev decides
- Probabilities, categories, or scores
- Code executes
- Home Assistant exposes answers as sensors and automation inputs
Inputs, sources & limits
Input · Home entity states and configured questions
It provides decision signals; automation rules control real devices. No real home hardware was tested here.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
Foadsf
Apply structured decisions to engineering workflows
- Jev decides
- Task routing, anomaly detection, and part matching
- Code executes
- Python rules execute the corresponding workflow
Inputs, sources & limits
Input · CAD, simulation logs, BOMs, and change information
Eight minimal examples explore workflows. They are not engineering-safety or manufacturing-quality certification.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
unownone
Check resumes and job fit with structured scores
- Jev decides
- Wording, structure, parseability, and job fit
- Code executes
- A Worker combines scores and produces interface copy
Inputs, sources & limits
Input · Resume text and an optional role profile
Without a key it uses deterministic mocks. Scores do not predict hiring outcomes. No real recruiting validation was performed here.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
Waxmell114514
Test trading decisions in a cost-aware simulator
- Jev decides
- Buy/sell direction and related judgments
- Code executes
- A matching simulator accounts for latency and fees
Inputs, sources & limits
Input · BTC/ETH prices, volume, and derived features
Simulated trading is not real order execution. Backtests, synthetic markets, and author charts do not establish realizable profits.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
luxus
Route smart-home voice requests
- Jev decides
- Whether the intent is a supported lighting command
- Code executes
- Simple commands call lighting services; others go to a Grok assistant
Inputs, sources & limits
Input · Home-control or chat utterances
The current fast path covers lighting services only. It does not establish support for all Home Assistant devices.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
uezo
Decide whether a speaker has finished
- Jev decides
- Whether to keep the user’s speaking turn open
- Code executes
- Map probabilities to wait times or turn completion
Inputs, sources & limits
Input · Speech transcripts and conversation context
Source provides a configurable turn-end gate. Conversation quality across languages and acoustic environments has not been established here.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
jarrodwatts
Choose trade direction from an on-chain order book
- Jev decides
- Buy or sell direction
- Code executes
- Order logic submits and replaces limit orders on Monad
Inputs, sources & limits
Input · The Kuru MON-USDC order book and price state
MODEL=mock is the default; no private key means dry-run. The public deployment labels fills as simulated. Not every run is a real Jev-driven trade.
Sources checked 2026-09-19 · Not independently run here
Public demoPublic post only
@verbove
Recommend nearby makers to meet
- Jev decides
- Maker matching; questions and choices are not disclosed
- Code executes
- The app shows nearby-maker recommendations; internals are undisclosed
Inputs, sources & limits
Input · Nearby maker profiles; exact input structure is undisclosed
The author claims accurate matching and shows results. No public source was verified; one example does not establish recommendation accuracy.
Sources checked 2026-09-19 · Not independently run here
Public demoPublic post only
Batch email classification experiment
@rileybrown
Explore batch inbox classification
- Jev decides
- Email categories
- Code executes
- The author shows batch classifications; the pipeline is undisclosed
Inputs, sources & limits
Input · Email text; samples and category labels are undisclosed
The post claims 500 emails in seconds for 3.5 cents. Reproducible tests, accuracy, and complete inputs are missing; these numbers are not endorsed here.
Sources checked 2026-09-19 · Not independently run here
Public demoPublic post only
Personal email client
@nickfthedev
Organize a private inbox with custom filters
- Jev decides
- Email ranking and filtering judgments
- Code executes
- The author’s private Mac/Linux client presents the inbox
Inputs, sources & limits
Input · Inbox messages and user-defined filters
Only the author’s post was found, not public source. Classification accuracy, distribution status, and the models behind other AI features could not be verified.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
itsmostafa
Give coding assistants structured decisions
- Jev decides
- Choice, Score, and Noul evaluations
- Code executes
- MCP returns typed results to the client
Inputs, sources & limits
Input · State and structured questions
Integration instructions were reviewed. Client installation and end-to-end latency have not been independently tested here.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
sharziki
Put semantic decisions into Unix pipelines
- Jev decides
- Classification, conditions, and scores
- Code executes
- Local thresholds produce filtered output and exit codes
Inputs, sources & limits
Input · Text, JSONL, or standard input
The project documents CLI and CI use. Example judgments do not establish accuracy on your own business data.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
markjaquith
Explore privacy and tone checks in a terminal
- Jev decides
- PHI, comment quality, tone, and categories
- Code executes
- A Rust CLI displays structured results
Inputs, sources & limits
Input · Text files, standard input, or short tasks
Includes privacy-detection experiments, not a medical compliance guarantee. Full input may be sent to the service. Not run independently here.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
tumf
Ask three kinds of structured questions from a terminal
- Jev decides
- Options, numeric scores, and conditions
- Code executes
- A Python CLI outputs JSON, primary values, or error codes
Inputs, sources & limits
Input · Text, JSON, files, or standard input
A community client requiring valid service credentials. Its existence does not establish that every provider was tested here.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
molis-ai
Package decisions as versioned services
- Jev decides
- Structured judgments under a fixed version
- Code executes
- Expose pinned functions through HTTP and MCP
Inputs, sources & limits
Input · Function definitions, state, and trial samples
The no-key demo explicitly uses simulated answers. A working interface does not prove a real Jev request was made.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
vercel
Integrate batched decisions through AI SDK
- Jev decides
- The TypeSafe evaluation model returns structured answers
- Code executes
- The SDK provides a unified evaluate call and types
Inputs, sources & limits
Input · Shared state and Choice, Score, or Boolean questions
Source confirms an optional TypeSafe provider. Evaluation is experimental; this does not mean all framework users use Jev by default.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
pydantic
Configure a TypeSafe provider in Python apps
- Jev decides
- Send evaluation requests to the Jev service
- Code executes
- The framework adapter manages clients, endpoints, and model configuration
Inputs, sources & limits
Input · App configuration and a TypeSafe SDK client
The review covers optional provider source code. It does not imply that all of Pydantic AI uses Jev.
Sources checked 2026-09-19 · Not independently run here
Docs / source reviewedSource available
Vicente-MD
Detect semantic failures inside HTTP 200 responses
- Jev decides
- Whether the body actually contains errors or maintenance notices
- Code executes
- Spring WebFlux raises normal exceptions for existing handlers
Inputs, sources & limits
Input · Service response bodies
The system fails open on API failure. Asynchronous calls still add latency and cost.
Sources checked 2026-09-19 · Not independently run here
Author testSource available
TokenTrim
Attribute failures in multi-agent traces
- Jev decides
- Responsible agent, key step, and error category
- Code executes
- The official scorer evaluates predictions
Inputs, sources & limits
Input · Failure traces and candidate agents, steps, and error types
The author uses an injected-error dataset. Some comparisons come from papers; constrained Who/When choices are not equivalent to free generation.
Sources checked 2026-09-19 · Not independently run here
Author testSource available
anessbelbati
Test Jev as a search reranker
- Jev decides
- Which passages are relevant and their order
- Code executes
- Evaluation code computes retrieval metrics and exposes raw output
Inputs, sources & limits
Input · Candidate passages returned by BM25
The repository compares multiple datasets. Different averaging methods can change rankings; small average differences do not establish an overall winner.
Sources checked 2026-09-19 · Not independently run here
Author testSource available
Gaurav-Gosain
Evaluate injection and vulnerable-code detection
- Jev decides
- Whether content is an injection or poses a security risk
- Code executes
- A Go evaluator computes metrics and saves per-sample results
Inputs, sources & limits
Input · Public-corpus messages and paired code samples
The author records 662 injection samples and 200 code pairs. Adding business context can change results; this is not a production-protection guarantee.
Sources checked 2026-09-19 · Not independently run here
No matching projects. Try a shorter query or reset the filters.