costwise-mcp: Instructions file for Codex

AGENTS.md

costwise-mcp AGENTS.md is an instructions file for Codex, OpenCode from okyashgajjar/costwise-mcp. It costs 3,635 tokens per session, scanned B, original, MIT.

Development instructions for Costwise MCP, a tool that helps AI agents work with repository information while reducing the amount of repeated context sent to language models. Its current work includes compact summaries, durable facts, and context-reducing tools.

In plain words
What is it for?
They help developers implement and test compact repository summaries, saved facts, and other tools for retrieving only the project information an agent needs.
Why use it?
They explain the project’s effort to lower prompt-cache costs by keeping unnecessary repository details out of the context used on each request.

Instructions file for CodexOpenCode

Written for Codex and OpenCode: the file is AGENTS.md. Also seen: reads .claude/ paths; positional $N argument; mentions Claude Code.

This is okyashgajjar/costwise-mcp's own configuration. It tells Codex and OpenCode how to work on costwise-mcp itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything costwise-mcp configures →

Not installable: its command points at a path on the author’s own machine, so it runs nowhere else. The line is /home/mryg/Research-Architectures/CLI/.

Reuse

Borrowing it

Nothing to install: this file belongs to okyashgajjar/costwise-mcp. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/okyashgajjar/costwise-mcp/main/AGENTS.md
Clone the repo
git clone --depth 1 https://github.com/okyashgajjar/costwise-mcp

Made for: Codex, OpenCode.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for costwise-mcp AGENTS.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/okyashgajjar/costwise-mcp/agents-md/github.svg)](https://agentmods.dev/instructions/okyashgajjar/costwise-mcp/agents-md)
Your own site
<a href="https://agentmods.dev/instructions/okyashgajjar/costwise-mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/okyashgajjar/costwise-mcp/agents-md/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for costwise-mcp AGENTS.md

Your own site · 80×15
<a href="https://agentmods.dev/instructions/okyashgajjar/costwise-mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/okyashgajjar/costwise-mcp/agents-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 3,635 This file is loaded in full into every session.
When invoked 3,635 The same file — it is already loaded in full.
Security scan B 1 finding. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What 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.

ModelPer sessionOnce invoked
Fable 5.1 $0.03635 $0.03635
Opus 5 $0.01818 $0.01818
Sonnet 5 $0.00727 $0.00727
Haiku 4.5 $0.00364 $0.00364

Measured 10d ago against content hash efa72beb7b4c, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade B, and why

costwise-mcp AGENTS.md scanned grade B with 1 finding 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 10d 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.

Asks for rootmediumPrivilege escalation

A mod that escalates privileges can change anything on the machine, not only the project.

On Ubuntu/Debian: `sudo apt install gcc libsqlite3-dev`
AGENTS.md · 218 lines

How it starts

The opening of the file, as written. The whole thing — 218 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Context for AI Coding Agents

Last updated by big-pickle on 2026-06-10.

V2 Foundation (in progress — branch feat/v2-foundation)

Goal: cut the dominant cost of long single sessions — Anthropic prompt-cache write/read, not model output. Evidence: a single call charged $2.95 where $2.84 was a 5-min cache write of ~455k tokens; output was only ~3.9k. The MCP cannot control when/how the client caches (breakpoints/TTL are client-owned); its only lever is reducing how many tokens ever enter the resident context window (the thing re-cached every turn). Every V2 piece serves that one goal.

Three steps:

  1. DONE — compact repo_summary. BuildRepositorySummaryCompact(ks, budget, module) in internal/retrieval/repository_summary.go: token-budgeted (reuses parseBudget), top modules by symbol count + +N more rollup, dropped the unbounded Layers chain, optional module drill-down. Tool gained budget+module params. Legacy BuildRepositorySummary/Format() untouched. (50-module synthetic: 1192→216 tokens; capped regardless of repo size.)
  2. DONE (code) — 3 cache-reducing tools in internal/mcpserver/tools.go: remember(repo_path,key,fact) (durable fact → kmemory UserNote + per-repo session_facts.json), stash_context(repo_path,content,label?) (park large blob out of window → tiny handle; file-backed internal/stash at <repoRoot>/.mycli-fts/stash/), recall(repo_path,query,source?,budget?) (query-scoped read of a stash by handle, or facts; hard-capped via step-1 budgeting). Stores wired into RepoSession (Stash, FactsPath, RememberFact, RecallFacts). Chosen over compact/summarize/forget because the user requires no context drop — stash is lossless (relocates tokens, re-fetchable). All tool outputs stay tiny. New names added to claude.go allow-list. Tests: internal/stash, internal/session/repo_session_v2_test.go.
  3. DONE — costwise-session skill (session-awareness). Teaches the model to keep the session lean (route large content through stash/recall, remember durable facts, prefer narrow retrieval). Single embedded source of truth: internal/skill/policy.md (go:embed, ~275 tok). Delivered two ways: (a) automatic/cross-IDE via server.WithInstructions(skill.Instructions()) in internal/mcpserver/server.go — every MCP client auto-loads it, zero install; (b) native Claude Code SKILL.md via costwise skill {install,uninstall,print} (cmd/skill.go) writing ~/.claude/skills/costwise-session/SKILL.md (or .claude/... with --local). install writes the skill by default (opt out --no-skill); uninstall removes it. Other IDEs rely on the instructions field + skill print for manual placement. internal/skill is standalone (NOT in the Target interface). Tests in internal/skill/skill_test.go.

Read the full file on GitHub · 218 lines

Changes

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.

  1. 10d ago First seen · 218 lines · 3,635 tokens per session scan B efa72beb7b4c

Subscribe to this mod's changes

costwise-mcp AGENTS.md is an instructions file published in the GitHub repository okyashgajjar/costwise-mcp (21 stars, last pushed 1mo ago), licensed MIT. It adds 3,635 tokens to every session, about $0.0182 per session on Opus 5. A static security scan graded it B with 1 finding (asks for root). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other instructions, from other repositories

data-scientist copilot-instructions.md

Repository instructions for a data-science workflow in GitHub Copilot, covering datasets, statistics, and analysis files.

realnghon/data-scientist · 373 tokens

next.js AGENTS.md

AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.

vercel/next.js · 7,296 tokens

codex AGENTS.md

AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.

openai/codex · 5,153 tokens

vscode buildNext.instructions.md

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 6,785 tokens

vscode oss-third-party-notices.instructions.md

Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).

microsoft/vscode · 5,001 tokens

langchain AGENTS.md

AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.

langchain-ai/langchain · 4,469 tokens