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/joaolfelicio/context-scribe/agents-mdgit clone --depth 1 https://github.com/Joaolfelicio/context-scribeWhat 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 | $0.00550 | $0.00550 |
| Opus 5 | $0.00275 | $0.00275 |
| Sonnet 5 | $0.00110 | $0.00110 |
| Haiku 4.5 | $0.00055 | $0.00055 |
Grade A, and why
context-scribe 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 2d 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 — 36 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
1. Project Goal
Context-Scribe is a background daemon that monitors AI logs to extract persistent rules and stores them in an MCP-managed Memory Bank. It manages a Retrieval-Based memory system where agents are instructed to actively fetch their context.
2. Agent Personas
- The Scribe: Silent observer. Monitors
~/.gemini/tmp/for JSON log updates. - The Evaluator: Decision engine. Identifies new rules and performs Autonomous Conflict Resolution (New rules overwrite old ones in the bank).
3. Global Bootstrap Logic (User Level)
To ensure all sessions across all directories are "Memory-Aware," the tool ensures the following instruction exists in ~/.gemini/GEMINI.md (or the equivalent rule file for the specific tool):
Master Retrieval Directive: "You have access to a persistent Memory Bank via the
memory-bank-mcp-server. Before beginning any task, you MUST invoke the appropriate tool (e.g.,read_memory_bank) to identify current project constraints and user preferences. Do not assume you have full context until this check is complete."
4. Conflict Handling (State Management)
The system operates on a Latest-is-Truth model:
- Detection: Scribe detects a new instruction in logs.
- Comparison: Evaluator checks the Bank for contradictory rules.
- Resolution: If a conflict is found, the new rule replaces the old one.
- Log: The tool outputs
[RESOLVE] Overwriting stale rule with new user preference.
5. Observability (The Heartbeat)
Real-time console feedback for background operations:
| Event | Level | Output Example |
|---|---|---|
| Detection | INFO |
[WATCH] Activity in ~/.gemini/tmp/chats/logs.json |
| Analysis | INFO |
[THINK] Evaluating last exchange for persistent rules... |
| Conflict | WARN |
[RESOLVE] New instruction 'Use Python 3.12' overwriting 'Use Python 3.10' |
| Commit | ACTION |
[BANK] Update successful via MCP tool 'write_file' |
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.
- 2d ago First seen · 36 lines · 550 tokens per session scan A c18a6be2e935
context-scribe AGENTS.md is an instructions file published in the GitHub repository Joaolfelicio/context-scribe (1 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 550 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
ClawRouter CLAUDE.md
Claude Code instructions for BlockRunAI/ClawRouter, covering clawrouter, commands, project structure, key dependencies and conventions.
pueue AGENTS.md
Instructions for Nukesor/pueue: This project does not approve of large-scale work done by LLMs.
SearChat CLAUDE.md
Claude Code instructions for yokingma/SearChat, covering claude.md, project overview, architecture, monorepo structure and key technologies.
InvestSkill GEMINI.md
Instructions for yennanliu/InvestSkill, covering investskill — gemini cli setup & usage guide, installation & setup, quick start, navigate to the investskill directory and start gemini cli (loads gemini.md automatically).
DevoxxGenieIDEAPlugin GEMINI.md
Instructions for devoxx/DevoxxGenieIDEAPlugin, a project described as: DevoxxGenie is an agentic plugin for IntelliJ IDEA that uses local LLM's (Ollama, LMStudio, GPT4All, Jan and Llama.cpp) and Cloud based LLMs to help review, test, explain your project code. Latest version now also supports Spec Driven Development…
aisix CLAUDE.md
Instructions for api7/aisix, covering claude.md, 1. think before coding, 2. simplicity first, 3. surgical changes and 4. goal-driven execution.