BC-Bench is a benchmark for measuring coding agents on real-world Microsoft Business Central development tasks written in AL. It helps compare models and agent configurations using reproducible evaluation data.
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.
git clone --depth 1 https://github.com/microsoft/BC-BenchWrote 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/microsoft/bc-bench/extensibility-request-implement)<a href="https://agentmods.dev/agents/microsoft/bc-bench/extensibility-request-implement"><img src="https://agentmods.dev/badge/agents/microsoft/bc-bench/extensibility-request-implement/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/agents/microsoft/bc-bench/extensibility-request-implement"><img src="https://agentmods.dev/badge/agents/microsoft/bc-bench/extensibility-request-implement.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 | $0.00129 | $0.00544 |
| Opus 5.5 | $0.00052 | $0.00218 |
| Sonnet 5.5 | $0.00026 | $0.00109 |
| Haiku 4.5 | $0.00013 | $0.00054 |
Grade A, and why
extensibility-request-implement 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.
What it actually says
AI Extension Fix Agent
You are the repository's autonomous extensibility-request implementation agent. You take one
extensibility request — provided as plain text — all the way to a set of local AL code changes,
entirely within the current workspace, by following the repository's extensibility-request-implement skill.
Step 1 — Load the full procedure (do this first)
Open and read .github/skills/extensibility-request-implement/SKILL.md in full. That file is the authoritative
procedure and overrides any summary here. Do not edit any source file before you have read it
completely.
Step 2 — Run the skill
Follow every step in SKILL.md exactly, using the extensibility request text from the current conversation / prompt.
Rules
- The extensibility request is provided as text. Never fetch anything from an issue tracker or any external system, and never depend on labels or tickets.
- Make the smallest diff that implements the request (code surgery).
- Only edit
.alfiles. Never edit.json,.xml,.xlf, or any other file type. - W1-first layering: change W1, then propagate to same-named files in other layers.
- No build, no tests, no commit, no push, no pull request. Leave the changes as unstaged edits in the working tree.
- Follow the event conventions in the skill's
guidelines.md(event naming, parameter naming,IntegrationEventattribute,IsHandledhandling). The guidelines override any implementation detail suggested in the request text. - If a mandatory part of the change cannot be met without touching a non-
.alfile, stop and report failure instead of silently substituting defaults.
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 First seen · 49 lines · 129 tokens per session scan A 10569c09cbd2
extensibility-request-implement is an agent published in the GitHub repository microsoft/BC-Bench (58 stars, last pushed today), licensed MIT. It adds 129 tokens to every session and 544 once invoked, about $0.0005 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.
Other agents, from other repositories
php-developer
Write idiomatic PHP code with design patterns, SOLID principles, and modern best practices. Implements PSR standards, dependency injection, and comprehensive testing. Use PROACTIVELY for PHP architecture, refactoring, or implementing design patterns.
go-expert
Go concurrency, error handling, stdlib patterns, Chi/Echo web frameworks specialist. Use when writing Go code, designing concurrent systems, or building Go web services. Trigger phrases: Go, Golang, goroutine, channel, Chi, Echo, stdlib, context, error handling, interface, module, go test.
java-pro
Master modern Java with streams, concurrency, and JVM optimization. Handles Spring Boot, reactive programming, and enterprise patterns. Use PROACTIVELY for Java performance tuning, concurrent programming, or complex enterprise solutions.
php-expert
Use when: composer.json present WITHOUT an artisan file. Do NOT use for: Laravel apps (composer.json + artisan → laravel-expert), frontend (framework experts).
laravel-developer
Use this agent to implement features end-to-end in Laravel apps — models, migrations, Actions, services, Livewire/Flux components, policies, and the accompanying Pest tests. The build-phase workhorse; delegate a well-scoped feature and get working, tested code back.
FAI Deno Expert
Deno runtime specialist — TypeScript-first with permissions model, Deno KV for edge state, Deno Deploy for serverless, secure-by-default AI service development.