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.
npx skills add marcellourbani/vscode_abap_remote_fs --skill repository-workflowgit clone --depth 1 https://github.com/marcellourbani/vscode_abap_remote_fsWrote 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/skills/marcellourbani/vscode_abap_remote_fs/repository-workflow)<a href="https://agentmods.dev/skills/marcellourbani/vscode_abap_remote_fs/repository-workflow"><img src="https://agentmods.dev/badge/skills/marcellourbani/vscode_abap_remote_fs/repository-workflow/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/skills/marcellourbani/vscode_abap_remote_fs/repository-workflow"><img src="https://agentmods.dev/badge/skills/marcellourbani/vscode_abap_remote_fs/repository-workflow.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 | $0.00071 | $0.04788 |
| Opus 5.5 | $0.00028 | $0.01915 |
| Sonnet 5.5 | $0.00014 | $0.00958 |
| Haiku 4.5 | $0.00007 | $0.00479 |
Grade A, and why
repository-workflow 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 22d 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 — 386 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ABAP Repository Comparison Workflow
Operate persistent, checkpointed repository comparisons between two connected SAP systems through the abapfs_*repository_workflow* LM tools.
Mental Model
The workflow has three user-visible stages:
- Scope and discovery finds matching TADIR objects independently on source and target, then automatically compares the two persisted inventories locally.
- Compare systems presents inventory presence, selects comparable objects, downloads both source snapshots, and compares their source locally.
- Assisted apply lets Copilot prepare a safety-reviewed source-to-target plan. The webview can stage one reviewed source file in a dirty target editor, but it never saves or activates SAP objects.
The source/target direction never changes after creation. Always state it explicitly before assisted-apply work.
Critical Operating Principle
This is a potentially high-impact, high-volume feature. Broad discovery can send many ADT requests to SAP; source comparison downloads objects from two systems; assisted apply can place source text in a live target editor. A mistaken scope, direction, selection, or concurrency choice can create substantial SAP load or prepare changes for the wrong system.
Copilot is the user's workflow assistant, not an autonomous repository scanner or change-application operator. The user sets the goal and guides progression. Copilot explains choices, prepares requested inputs, runs only explicitly requested stages, reports results, and stops at decision boundaries.
Mandatory behavior:
- Never create, configure, run, resume, rerun, or prepare an assisted-apply plan merely because it appears to be the next workflow step.
- Require clear user direction before every potentially expensive stage: criteria update, discovery with its automatic local inventory comparison, source comparison/download, and assisted-apply plan preparation.
- Do not interpret “inspect,” “explain,” “review,” “check,” or “what happens?” as permission to run or mutate a workflow.
- Treat discovery and its local inventory comparison as one requested operation. Do not automatically continue from that operation into source comparison or assisted apply unless the user explicitly requested the complete sequence.
- If the user requested a multi-stage sequence, stop immediately if they pause, redirect, or change scope. The newest instruction controls.
- Before discovery, restate source, target, effective scope, subpackage behavior, and any unusually broad pattern. Ask for confirmation when scope or intent is ambiguous.
- Before source comparison, state how many objects will be selected, the source/target SAP download concurrency, and the local verification concurrency. Read the existence summary or bounded artifact first when the count is unknown.
- Before assisted-apply plan preparation, state source-to-target direction and make clear that planning is non-mutating. Prepare it only when requested.
- Copilot's role ends after preparing and explaining the plan. Only the user can stage a file from the webview, save it through the normal editor flow, choose a transport, and activate related objects.
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.
- 22d ago First seen · 386 lines · 71 tokens per session scan A 369e2782b96b
repository-workflow is a skill published in the GitHub repository marcellourbani/vscode_abap_remote_fs (399 stars, last pushed 4d ago), licensed MIT. It adds 71 tokens to every session and 4,788 once invoked, about $0.0003 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 2026-09-24.
Other skills, from other repositories
oh-my-posh
Install, configure, or troubleshoot Oh My Posh/ohmyposh: shell init, themes, segments, Nerd Font icons, and prompt setup on PowerShell, zsh, bash, or fish.
code-changes
Workflow for any task that ends in code changes: issue analysis or triage, pull request review comments, features, bug fixes, refactors. Invoke at the start of such a task, before reading or writing code.
mma-investigator
Expert system for investigating MMA (Multi-Metric Allocator) behavior on CockroachDB clusters. Helps oncall engineers diagnose load imbalances, understand rebalancing decisions, and identify why MMA did or didn't act.
zdx-investigate-multi-app-outage
Diagnose a multi-application outage scoped to one location by correlating ZDX alerts, affected devices, and shared cloud-path hops. Identifies the devices affected at a specific office, compares the per-application network path across multiple SaaS apps to surface the common network bottleneck, and produces an…
zdx-diagnose-deeptrace
Run a ZDX deep trace diagnostics session to investigate network and device issues. Start new sessions, analyze web probe metrics, cloud path topology, device health, top processes, and event timelines to pinpoint root cause. Use when an administrator asks: 'Start a deep trace for this user', 'Analyze the diagnostics…
zdx-investigate-alerts
Investigate active and historical ZDX alerts to understand their scope, root cause, and impact. Drills into affected devices, correlates with application metrics, and identifies patterns across time. Aligned with ZDX Copilot troubleshooting use cases. Use when an administrator asks: 'Show me ongoing alerts', 'What…