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/gru-953/mta_plugin/claude-mdgit clone --depth 1 https://github.com/GRU-953/MTA_PluginWrote 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.
[](https://agentmods.dev/instructions/gru-953/mta_plugin/claude-md)<a href="https://agentmods.dev/instructions/gru-953/mta_plugin/claude-md"><img src="https://agentmods.dev/badge/instructions/gru-953/mta_plugin/claude-md.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00739 | $0.00739 |
| Opus 5 | $0.00369 | $0.00369 |
| Sonnet 5 | $0.00148 | $0.00148 |
| Haiku 4.5 | $0.00074 | $0.00074 |
Grade A, and why
MTA_Plugin CLAUDE.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 3d 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 — 41 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md — Memorised them All
This repo is under a multi-session hardening/extension program. Cross-session
state lives in program/ on the develop branch — that is the single source of
truth, not model memory. Any fresh session resumes from it with zero prior context.
Resume protocol (every session)
git checkout develop && git pull.- Read
program/PROGRESS.md— the ▶ RESUME HERE pointer + the WP table — and the last 1–2 entries ofprogram/SESSION_LOG.md. SkimDECISIONS.md/RISKS.md/ACCEPTANCE.mdas needed. Load only the code the next Work Package touches — do not re-read the whole tree. - Do the next unblocked WP on a
wp-<id>-<slug>branch → PR intodevelopwith green CI. Never pushmaindirectly; release frommainvia avX.Y.Ztag (the tag-triggered train publishes all channels — seeprogram/PUBLISH_MANIFEST.md). - Wrap up: leave the repo green + resumable; update
PROGRESS.md(statuses + ▶ RESUME HERE) and append aSESSION_LOG.mdentry ending with the single exact next step.
Status
v2.6.1 SHIPPED — PyPI + GitHub Release + .mcpb + Homebrew tap + GHCR (OIDC trusted-publish,
SBOM, cosign single-file *.sigstore.json bundle). Ships v2.5.0 cross-AI mta setup
(Claude/Gemini/Cursor/VS Code/Windsurf/Codex; Grok auto-discover) + v2.6.0 recall/resolve perf +
supply-chain. v2.6.1 = the v2.6.0 package with two release-pipeline fixes (SBOM attach + cosign pin).
S25 docs reframe SHIPPED (no version bump): README + all 7 metadata surfaces reframed
Claude-only → cross-AI; live on GitHub / Claude-marketplace / Glama (PR #80→main, #81→develop).
Next = v2.x+ backlog (optional): a docs-only v2.6.2 to refresh PyPI's page with the new
cross-AI description; remote-MCP recipe polish; the deferred Low items in program/RISKS.md
(R-16/R-17, accepted).
Invariants (must never regress)
Token-free (tiny tool results; document contents never returned to the model) · 100% local / no telemetry · dependency-free classical/offline fallback (a digest succeeds with no models and no network) · atomic, crash-safe writes. CI must stay green on the Ubuntu/macOS/Windows × 3.10/3.12 matrix.
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.
- 3d ago First seen · 41 lines · 739 tokens per session scan A 272b8b2b3515
MTA_Plugin CLAUDE.md is an instructions file published in the GitHub repository GRU-953/MTA_Plugin (1 stars, last pushed 29d ago), licensed MIT. It adds 739 tokens to every session, about $0.0037 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
cosmos-mcp CLAUDE.md
Instructions for teampolarity/cosmos-mcp, covering cosmos-mcp, layout, tools (authoritative list in src/tools/index.ts), cli subcommands and imessage is the headline source.
vibe-mcp AGENTS.md
Instructions for VibeTechnologies/vibe-mcp, covering vibe-mcp agent memory, canonical eval reference, relay and connection semantics, what counts as real e2e and fake extension test harness rules.
delx-mcp-server AGENTS.md
AGENTS.md instructions for davidmosiah/delx-mcp-server, covering agent development notes, scope, commands, rules and agent-readiness checklist.
codedb AGENTS.md
AGENTS.md instructions for justrach/codedb, covering codedb agent guidelines, what codedb is (and isn't), review guidelines, pre-merge verification and security-sensitive areas.
repo-context-mcp AGENTS.md
Instructions for nduc99911/repo-context-mcp, covering agents.md — repo-context-mcp, project, layout, commands and conventions.
goodboy AGENTS.md
AGENTS.md instructions for akhayam99/goodboy, covering agents.md, where new code goes, naming, components and exports and store selectors and memoization.