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/addicted2crypto/dcgp/agents-mdgit clone --depth 1 https://github.com/addicted2crypto/DCGPWhat 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 | $0.03449 | $0.03449 |
| Opus 5 | $0.01724 | $0.01724 |
| Sonnet 5 | $0.00690 | $0.00690 |
| Haiku 4.5 | $0.00345 | $0.00345 |
Grade A, and why
DCGP 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 — 256 lines — stays where its author put it; the contents beside it link to each section on GitHub.
DCGP - Agent Mounting Guide
Dynamic Context Guidance Paths is the semantic operating system for LLM agents. This file is the single source of truth for any agent operating in this repository.
Modern AI tools that support AGENTS.md (Claude Code, Cursor, Zed, OpenAI Codex CLI, and a growing list) read this file directly. Tools that insist on their own filename can be aliased in one line (cp AGENTS.md .clinerules etc.) - DCGP itself ships only this file plus HARDRULES.md and CLAUDE.md to keep the governance surface slim. For runtime tool access rather than static file reads, use @dcgp/mcp (see README).
Live implementation state is defined by ./scripts/verify-dcgp.sh. If this document claims a feature the script does not verify, the script wins. Re-read this file only after verify-dcgp.sh exits 0.
Conformance tier claimed by this repo
DCGP-1.0-FULL + EXTENDED
Phase A ships the governance kernel plus the FineTuningExporter. Phases B-D (16 community paths, OpenCode plugin, CLI, VS Code extension) are explicitly deferred and their absence is not a bug.
What this system does
DCGP prevents context decay in long LLM sessions by running a closed-loop control system around the model. It is not a context pruner. It is not a summarizer. It is a governance layer - a kernel that manages what the model knows, how certain it is about its domain, when to intervene before hallucinations begin, and how much of the working context survives pruning.
Every agent operating in this codebase operates inside this loop. There are no exceptions.
The 7-step Circular Intelligence Loop
Sense -> Classify -> Predict -> Orchestrate -> Execute -> Verify -> Refine -> (back to Sense)
Step 1 - Sense (FingerprintEngine): Reads the workspace before you type. Parses packages, env vars, git branch, config files. 30s TTL cache. 16 ignored dirs. Zero shell calls.
Step 2 - Classify (DomainClassifier): Scores all registered domain paths by weighted signal match. Confidence 0.0-1.0. Signal weights: packages(0.95) > files(0.90) = gitBranch(0.90) > env(0.85) > tools(0.75) > keywords(0.60). Session keywords capped at 500. Detects collision (top-two confidences within COLLISION_DELTA = 0.10) and deadlock (A->B->A within SHIFT_COOLDOWN_TURNS = 3). Forced reclassification every CLASSIFIER_TTL_TURNS = 20 turns regardless of entropy state (stale-classifier blind spot).
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 · 256 lines · 3,449 tokens per session scan A ea42b9c5ee51
DCGP AGENTS.md is an instructions file published in the GitHub repository addicted2crypto/DCGP (5 stars, last pushed 3d ago), licensed MIT. It adds 3,449 tokens to every session, about $0.0172 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
plannotator AGENTS.md
Instructions for backnotprop/plannotator, covering plannotator, project structure, server runtimes, installation and environment variables.
codex-autoresearch AGENTS.md
Instructions for TheGreenCedar/codex-autoresearch, covering agents.md, purpose and scope, canonical sources, stable product boundaries and source and package boundaries.
autoharness AGENTS.md
Instructions for tigerless-labs/autoharness: This repo uses Open Knowledge — collaborative markdown via MCP. .open-knowledge/config.yml (with optional /.open-knowledge/config.yml; CLI/env may override) is the path contract: content.dir is the root for relative paths; content.include lists globs that add markdown…
solo AGENTS.md
Instructions for solo-agent/solo, covering project testing rules and project service lifecycle rules.
superset AGENTS.md
AGENTS.md instructions for superset-sh/superset, covering superset monorepo, project structure, exception: shadcn/ui components, database and releases.
untether GEMINI.md
Instructions for littlebearapps/untether, covering untether — gemini instructions, stack & conventions, architecture, key rules and testing.