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 agentmods add commands/arc-mcp/arc-1/deep-issuegit clone --depth 1 https://github.com/arc-mcp/arc-1What 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 | $0.00000 | $0.02538 |
| Opus 5 | $0.00000 | $0.01269 |
| Sonnet 5 | $0.00000 | $0.00508 |
| Haiku 4.5 | $0.00000 | $0.00254 |
Grade A, and why
deep-issue 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 3d 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 — 94 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Deep Issue — research & validate an external GitHub issue against real sources
Deeply research a community-filed issue on arc-mcp/arc-1 before anyone touches code. The whole point: decide whether the issue is real, and why, from verified evidence — not from the reporter's claim and not from memory. ARC-1's worst bugs come from reasoning about ADT behavior instead of checking what the system actually does; the same trap turns a plausible-sounding issue into a wrong fix.
Use this for any non-trivial external issue (bug report, feature request, "X doesn't work on release Y"). For an obvious typo-fix or a duplicate you can close in one line, just do that directly.
The output is a validated dossier + a paste-able GitHub reply + a recommendation — not a code change. When the verdict is "fix it," hand the dossier to /deep-feature (which owns research→plan→implement→PR).
Input
An issue number or URL (e.g. 379 or https://github.com/arc-mcp/arc-1/issues/379). If the user gives a vague description instead, ask for the number, then start.
The prime directive
No verdict until you have the full picture. "Full picture" means you can answer, with evidence: is the reported behavior real on a live system, what the ADT endpoint actually does, whether HEAD already fixes it, what the true root cause is, and which releases are affected. Until then you are still in Phase 1.
A reporter saying "reproduced on SAP_BASIS 8.16" is a lead, not a fact — confirm it. A 200 OK does not mean a filter/type/parameter was honored; ADT silently ignores unknown ones. Verify response content, not just status. The issue may be already fixed on HEAD (cf. #378: the fix existed, only a stale comment + doc hint remained) or a duplicate of a known dossier.
Research Sources (concrete — same sources /deep-feature uses)
All ~/DEV/* repos are read-only references — never modify them. If a path is missing, note it and continue.
| Source | Where | Use it for |
|---|---|---|
| The issue itself | gh issue view <n> --comments; linked PRs via gh issue view <n> --json |
The claim, the repro steps, prior maintainer replies, any linked fix attempt |
| ARC-1 project guide | CLAUDE.md, INFRASTRUCTURE.md |
Conventions, Key-Files map, code patterns; the 3 live systems + how to reach them |
| Current HEAD code | the cited src/... files (use grep -a — intent.ts contains raw NULs) |
Is it already fixed / never was as described? Read the exact lines the issue cites before believing them |
| Prior issue dossiers | docs/research/issues/ (e.g. 293-ecc-423-invalid-lock-handle.md) |
Duplicate? Same root cause as a solved one? Reuse the proven analysis |
| ARC-1 live research notes | docs/research/, type-code evidence under docs/research/abap-types/ |
Prior live spikes & ground truth — read first so you don't re-derive proven facts |
| Competitor / reference trackers | docs/compare/ (esp. docs/compare/05-fr0ster-mcp-abap-adt.md, docs/compare/01-vibing-steampunk.md) |
Did a sibling client hit the same SAP bug? Issues often cite fr0ster/vibing-steampunk precedent |
| Eclipse ADT — apidoc + contracts | ~/DEV/arc-1-eclipse-adt/ (com.sap.adt.core.apidoc-*, api/01..20-*.md) |
Exact ADT endpoint URIs, media types, request/response contracts. Ground truth for "what should this endpoint do" |
| SAP ADT language server | ~/DEV/arc-1-lsp/ (vendor/adt-ls, docs/adt-ls-*) |
How SAP's own server calls ADT — a second independent witness |
| Reference ADT-over-MCP impls | ~/DEV/mcp-abap-adt/, ~/DEV/mcp-abap-adt-fr0ster/ |
How others implement the same op — and whether they fixed this exact bug |
| ABAP language reference | ~/DEV/abap-docs/docs/ |
ABAP keyword/syntax semantics when the issue touches language constructs |
| SAP docs / Notes (MCP) | sap-docs MCP (search, fetch, sap_community_search, sap_search_objects, abap_feature_matrix) + sap-notes MCP (search, fetch) |
Official docs, community threads, and SAP Notes/KBAs for known corrections — cite the Note number when a fix depends on it. For broad web research, /deep-research |
| Live SAP systems | arc1-cli call <Tool> ..., npm run probe -- --save-fixtures tests/fixtures/probe/<name>. Systems: NW 7.50 (npl), S/4HANA 2023 / 758 (a4h), ABAP Platform 2025 / 816 (a4h-2025) — creds + recipes in INFRASTRUCTURE.md |
Ground truth. Reproduce the issue yourself, on the release(s) it names |
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.
- 3d ago First seen · 94 lines · 0 tokens per session scan A 2c19ae6c02c4
deep-issue is a command published in the GitHub repository arc-mcp/arc-1 (174 stars, last pushed 2d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 2,538 tokens. 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 commands, from other repositories
git
Git operations with intelligent commit messages and workflow optimization.
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
analyze
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.