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/junmystery/agent-guidance-python/agents-mdgit clone --depth 1 https://github.com/JunMystery/Agent-Guidance-PythonWhat 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 | $0.01661 | $0.01661 |
| Opus 5 | $0.00830 | $0.00830 |
| Sonnet 5 | $0.00332 | $0.00332 |
| Haiku 4.5 | $0.00166 | $0.00166 |
Grade A, and why
Agent-Guidance-Python 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.
How it starts
The opening of the file, as written. The whole thing — 80 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Token Saving — Built-in Optimization
agent-guidance-mcp saves tokens through 4 automatic mechanisms:
- Bounded reads —
agent-guidance-mcp_project_context(operation="read")caps at 300 lines;searchcaps snippets at 300 chars andlimitat 20 results. Replaces expensive raw file reads. - Content compression — Language-aware comment/whitespace removal, dedup, badge-stripping. Applies to every read/search response automatically.
- Per-type token budgets — Source files 3K, docs 2K, skills 3K, task_pipeline 12K tokens max per call.
- Tracking & reporting — Every optimized call records original vs. optimized tokens to SQLite. View with
agent-guidance-mcp_token_stats(session) oragent-guidance-mcp_usage_report(scope="all")(lifetime).
Config (already set globally): AGENT_GUIDANCE_FILTER_LEVEL=aggressive, doc/skill caps reduced to 2K/3K — aggressive filtering by default.
How to maximize savings:
- Always use
agent-guidance-mcp_task_pipeline→agent-guidance-mcp_project_context/agent-guidance-mcp_guidanceinstead of raw tools (Rule 3). - Call
agent-guidance-mcp_token_statsat end of each phase to verify compression ratio. - If token usage is still high, set
AGENT_GUIDANCE_TOKEN_OPT=0to disable (not recommended).
Tool naming note: Some MCP hosts prefix tool names with agent-guidance-mcp_ (e.g. agent-guidance-mcp_task_pipeline), while others expose bare names (task_pipeline). The examples below use prefixed names. When calling, always use the exact tool name shown in your available-tools list — never add or drop a prefix.
CRITICAL — Tool Rules (READ FIRST)
For EVERY user interaction — planning, implementation, testing, debugging, reviewing, refactoring, or any other action:
Agent Guidance MCP — Tool Selection Priority
| You need to... | Use THIS tool first | Why |
|---|---|---|
| Start any task or phase | agent-guidance-mcp_task_pipeline(task="...") |
Recommendations + tree + code search + UI in ONE call |
| Check coding standards / skills | agent-guidance-mcp_guidance(operation="search", query="...") |
No other tool provides standards or skill lookup |
| Read a file | agent-guidance-mcp_project_context(operation="read", relative_path="...") |
Token-capped at 300 lines — prevents context blowout |
| Search codebase text | agent-guidance-mcp_project_context(operation="search", query="...") |
Ranked, bounded results. Fallback when codegraph unavailable |
| Understand code structure | agent-guidance-mcp_project_context(operation="structure", relative_path="...") |
Hierarchical view of classes, methods, functions in a file |
| Extract symbols | agent-guidance-mcp_project_context(operation="symbols", relative_path="...") |
Flat list of classes, functions, methods with signatures |
| Get structured workflow | agent-guidance-mcp_guidance(operation="workflow", identifier="plan") → "code" → "test" |
Enriched dev workflow with auto-chaining |
| Pre-code checklist | agent-guidance-mcp_guidance(operation="precode", query="...") |
Coding conventions, security, testing, arch, deploy rules |
| Post-code verification | agent-guidance-mcp_guidance(operation="verify", query="...") |
Auto-detect test/review/security/audit/deploy; suggests next workflow |
| Skill feedback loop | agent-guidance-mcp_guidance(operation="feedback", identifier="...", rating=1-5, query="...") |
Rate skills to boost future recommendations for similar tasks |
| Find symbol references | agent-guidance-mcp_project_context(operation="references", query="...") |
Locate all usages of a symbol across the codebase |
| Get UI/design guidance | agent-guidance-mcp_ui_ux(operation="search", query="...") |
Style, colors, typography, charts, slides |
| Check workflow stage | agent-guidance-mcp_workflow_gate(action="check"/"status"/"set_stage") |
Verify plan approvals and progress lifecycle |
| Persist/recover session | agent-guidance-mcp_session_continuity(operation="save"/"load"/"clear") |
State recovery / task checklist continuity |
| Browse project tree | agent-guidance-mcp_project_context(operation="tree") |
Optimized directory tree view |
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 · 80 lines · 1,661 tokens per session scan A 44ce41aada1c
Agent-Guidance-Python AGENTS.md is an instructions file published in the GitHub repository JunMystery/Agent-Guidance-Python (2 stars, last pushed 1mo ago), licensed MIT. It adds 1,661 tokens to every session, about $0.0083 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
guide copilot-instructions.md
Instructions for rios0rios0/guide, covering software development guide repository, working effectively, quick setup and validation, core build process and github actions workflows.
guide CLAUDE.md
Instructions for rios0rios0/guide, covering claude.md, project overview, build, test, and validate, update-wiki tool (go 1.26.2) — syncs docs to github wiki and toc sync validation (run from repo root).
constitution AGENTS.md
Instructions for systemfsoftware/constitution, covering agents.md — constitution repository, startup workflow, working rules, amending the constitution and file split.
android-finance-spec AGENTS.md
Instructions for brycegao/android-finance-spec: When generating, editing, reviewing, or refactoring this repository, follow all Android finance specs.
android-finance-spec CLAUDE.md
Instructions for brycegao/android-finance-spec: This repository packages Android finance coding specs as an installable AI skill. Before generating, editing, or reviewing code, read.
android-finance-spec copilot-instructions.md
Instructions for brycegao/android-finance-spec: When generating or editing Kotlin/Android docs, examples, scanner logic, or skill files, follow.