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 instructions/nikzdevz/obsmcp/agents-mdgit clone --depth 1 https://github.com/nikzdevz/OBSMCPWhat 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 | $0.00559 | $0.00559 |
| Opus 5 | $0.00280 | $0.00280 |
| Sonnet 5 | $0.00112 | $0.00112 |
| Haiku 4.5 | $0.00056 | $0.00056 |
Grade A, and why
OBSMCP AGENTS.md 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 yesterday.
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
AGENTS.md
You are operating in a shared multi-model continuity system backed by obsmcp.
Before doing substantive work, read in this order:
.context/PROJECT_CONTEXT.md.context/CURRENT_TASK.json.context/HANDOFF.md.context/DECISIONS.md.context/BLOCKERS.json.context/RELEVANT_FILES.json.context/SESSION_SUMMARY.md- Check if a Code Atlas exists for this project. Call
get_code_atlas_status()via MCP or runctx.bat atlas statusfrom the shell. If the atlas does not exist, callscan_codebase(). For large or first-time scans,scan_codebase()may return a background job instead of a finished atlas; pollget_scan_job()orwait_for_scan_job()until it completes. The Code Atlas documents every file, function, class, feature, and cross-reference in the project — it gives you a complete structural understanding without reading every source file. - For quick orientation: call
get_audit_log(limit=10)to see recent activity. - For sprint planning: use
bulk_task_opsto batch-create or batch-update tasks. - For token-efficient context: use
generate_compact_context_v2(max_tokens=3000)— includes decision chains, dependency map, session info, and recommended semantic lookups. - For low-latency startup or resumed work, use the cached tiered context and delta tools first:
generate_context_profile(profile="fast"|"balanced"|"deep"|"handoff"|"recovery")generate_delta_context(...)
- When you need targeted understanding instead of rereading large files, use semantic tools first:
describe_module(module_path)describe_symbol(symbol_name, module_path?)describe_feature(feature_name)search_code_knowledge(query)get_symbol_candidates(symbol_name)if a symbol name is ambiguousget_related_symbols(entity_key)to expand from one symbol to its neighbors
- For dependency overview: use
get_blocked_tasks()andvalidate_dependencies()to check task readiness.
Rules:
- Do not assume you are the first or only agent.
- Continue the current task instead of restarting discovery.
- Preserve continuity notes, blockers, and decisions.
- When you finish a meaningful chunk, log work to
obsmcp. - Before you stop, create a handoff for the next model or tool.
Preferred write paths:
- MCP tools if available
ctx.batif MCP is not available
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.
- yesterday First seen · 42 lines · 559 tokens per session scan A 675c369e3a16
OBSMCP AGENTS.md is an instructions file published in the GitHub repository nikzdevz/OBSMCP (0 stars, last pushed 4mo ago), licensed MIT. It adds 559 tokens to every session, about $0.0028 per session on Opus 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 2026-08-31.
Other instructions, from other repositories
mcpelevator CLAUDE.md
Instructions for pacnpal/mcpelevator, covering commands, architecture, conventions, agent skills and issue tracker.
mcpelevator AGENTS.md
Instructions for pacnpal/mcpelevator, covering commands, architecture and conventions.
context-loom AGENTS.md
AGENTS.md instructions for ownasquare/context-loom, covering context loom agent guide, product contract, architecture, security and tests.
maple-procure CLAUDE.md
Claude Code instructions for samarpassey/maple-procure, covering mapleprocure, commands, read these when and repo conventions.
1cent AGENTS.md
AGENTS.md instructions for maxzoa/1cent, a project described as: Pay-per-call URL intelligence for AI agents via MCP and x402 USDC on Base Mainnet.
mcp-sql-server AGENTS.md
AGENTS.md instructions for harryvaldez/mcp-sql-server, covering agents, project snapshot, fast start, architecture boundaries and non-negotiable guardrails.