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/cdeust/ai-architect-mcp-codebase/claude-mdgit clone --depth 1 https://github.com/cdeust/ai-architect-mcp-codebaseWrote 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/cdeust/ai-architect-mcp-codebase/claude-md)<a href="https://agentmods.dev/instructions/cdeust/ai-architect-mcp-codebase/claude-md"><img src="https://agentmods.dev/badge/instructions/cdeust/ai-architect-mcp-codebase/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 | $0.00613 | $0.00613 |
| Opus 5 | $0.00307 | $0.00307 |
| Sonnet 5 | $0.00123 | $0.00123 |
| Haiku 4.5 | $0.00061 | $0.00061 |
Grade A, and why
ai-architect-mcp-codebase 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 4d 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 — 39 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ai-architect-mcp-codebase
Rust MCP server: parses repositories with tree-sitter, builds a code graph, answers structural queries. See @CONTRIBUTING.md for the layer rules and coding standards.
Global rules — model behavior, the loop discipline, and the zetetic standards — are imported, not restated here:
@/.claude/rules/model-behavior.md
@/.claude/rules/coding-standards.md
Environment quirks — these will cost you a session if ignored
- This clone is live-mounted as the installed plugin.
~/.claude/plugins/cache/.../0.9.x/target/release/ai-architect-mcp-codebasesymlinks here. Nevercheckout,pull,stashor build a different branch in it — the running MCP server dies. Work in a worktree:git worktree add ../automatised-pipeline-wt-<topic> -b <branch> origin/main. - The launcher verifies the release binary's SHA-256 against a pin. A dev rebuild breaks it unless
AI_ARCHITECT_SOURCE_CHECKOUT=1is exported (it is, in~/.zshenv). A dead server shows only asMCP error -32000: Connection closed; the real message appears when runningbin/launch-plugin.shmanually withCLAUDE_PLUGIN_ROOTset. bin/ensure-binary.shpins the SHA-256 ofCargo.tomland.claude-plugin/plugin.json. Both drift whenever dependencies or version change — update them or CI fails.
Commands
cargo test --lib # unit tests
cargo test --test graph_accuracy # accuracy gate
cargo run --release -p bench-end-result --bin bench_end_result -- --all # note the -p
The bench archives to benches/runs/<ts>.md (gitignored; force-add only a release manifest).
Exit 0 requires aggregate ≥0.85 and every language ≥0.75.
Gates that fail CI but not local runs
- Doc-truth:
scripts/check_doc_claims.pyfails the build when README's test count or coverage badge drifts from measured reality (100-test buckets). Crossing a bucket means updating README and committing it. - clippy
--all-targetscompileslibseparately fromunittests: anything reachable only from#[cfg(test)]code, imports included, must itself be#[cfg(test)]-gated. - CI catches blast radius that local gates miss — activating a resolver path can break a cross-repo defense test. Watch the run, don't assume.
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.
- 4d ago First seen · 39 lines · 613 tokens per session scan A 264af7634bd9
ai-architect-mcp-codebase CLAUDE.md is an instructions file published in the GitHub repository cdeust/ai-architect-mcp-codebase (4 stars, last pushed today), licensed MIT. It adds 613 tokens to every session, about $0.0031 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
brooks-lint CLAUDE.md
Claude Code instructions for hyhmrright/brooks-lint, covering claude.md, what this repo is, harness: brooks-lint maintenance, workflow conventions and critical gotchas.
brooks-lint AGENTS.md
AGENTS.md instructions for hyhmrright/brooks-lint, covering agents.md, core purpose, skill integration, engineering standards and project structure.
lsp-tools copilot-instructions.md
Instructions for zircote/lsp-tools, covering copilot instructions, priorities, key files, commands and when adding language support.
lsp-tools CLAUDE.md
Instructions for zircote/lsp-tools, covering claude.md (project context), repo goals, key files, development conventions and when adding tools.
weld cpp.instructions.md
C++ naming and structure guidance.
weld testing.instructions.md
Test guidance for implementation changes.