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 skills add cdeust/ai-architect-mcp-codebase --skill understand-codebasegit 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/skills/cdeust/ai-architect-mcp-codebase/understand-codebase)<a href="https://agentmods.dev/skills/cdeust/ai-architect-mcp-codebase/understand-codebase"><img src="https://agentmods.dev/badge/skills/cdeust/ai-architect-mcp-codebase/understand-codebase/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.
<a href="https://agentmods.dev/skills/cdeust/ai-architect-mcp-codebase/understand-codebase"><img src="https://agentmods.dev/badge/skills/cdeust/ai-architect-mcp-codebase/understand-codebase.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00045 | $0.00560 |
| Opus 5.5 | $0.00018 | $0.00224 |
| Sonnet 5.5 | $0.00009 | $0.00112 |
| Haiku 4.5 | $0.00005 | $0.00056 |
Grade A, and why
understand-codebase 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 1mo 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.
Understand Codebase
Use the eight-tool core profile. Do not assume that pipeline-only tools are available.
For Rust caller or impact questions, request lsp: true in analyze_codebase
and make rust-analyzer available. Inspect lsp_status.state: disabled means
static-only analysis; failed means analysis continued and may retain partial
LSP results. Record the error and treat affected conclusions as unverified until
source checks or a successful rerun support them. completed means the pass
returned, not that every call resolved. Record unresolved sites and coverage gaps.
If a full-profile workflow supplies a check_security_gates receipt, inspect
both gates_passed and assessment_complete. A true pass with an incomplete
assessment is not security clearance: report skipped checks, unresolved symbols
and warnings. This core-profile workflow does not invoke that additional tool.
Workflow
- Call
health_checkand record the active profile and tool count. Stop with an installation or configuration instruction if the server is unavailable. - Resolve the repository root to an absolute path. Choose a task-scoped, absolute, writable
output_dirfor graph artifacts; never use the source root or a broad system directory as that output. - Call
analyze_codebasewith the repository path and output directory. Reuse the exactgraph_pathreturned by the tool in every later call. - Immediately call
query_graphwithgraph="missed". Treat parse-incomplete, skipped, and quarantined files as explicit coverage gaps. Absence from this report is not proof of complete indexing. - Use
search_codebaseto discover qualified symbol names. Drill into relevant hits withget_contextorget_symbol; do not guess qualified names. - Use read-only
query_graphqueries for structural questions the typed tools do not answer. AddORDER BYbefore paging withnext_offset. - For any important negative result, inspect the coverage report and use targeted source search for the relevant symbols and call forms, including unflagged files, before concluding that a symbol or edge does not exist. A clean coverage report does not establish relationship completeness.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 1mo ago Changed · +12 lines f64124469612
- 1mo ago First seen · 24 lines · 45 tokens per session scan A 5efece92fcee
understand-codebase is a skill published in the GitHub repository cdeust/ai-architect-mcp-codebase (7 stars, last pushed today), licensed MIT. It adds 45 tokens to every session and 560 once invoked, about $0.0002 per session on Opus 5.5. 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 skills, from other repositories
bug-fix
Investigate, reproduce, and safely fix a bug with regression protection. Composes context, diagnosis, architecture, code quality, and testing guardrails into a reproduce-first repair workflow. Use when the user says 'fix this bug', 'debug this', 'investigate this failure', 'patch this regression', 'repair this issue'…
refactor-safely
Restructure existing code safely without changing externally observable behavior. Composes context, design, architecture, code quality, and testing guardrails into a characterization-first refactoring workflow. Use when the user says 'refactor this', 'clean this up', 'untangle this module', 'move this to the right…
workflow-bug-triage
Use when investigating a bug to root-cause and FILE a GitHub issue rather than fixing immediately — counterpart to /swe-workbench:debug (which patches). Enforces the Iron Law (no patches without root cause), and produces a structured issue with code-path table and impact assessment.
workflow-audit-emit-issues
After a codebase audit, groups audit findings by subsystem (path prefix) and emits them as context-rich GitHub issues with template discovery, label selection, and a batch preview-confirm gate before filing. Keywords — audit findings grouped github issues subsystem emit.
workflow-codebase-audit
Use for cold-start, time-boxed, multi-axis audits of unfamiliar codebases — take-home assessments, post-acquisition or due-diligence reviews, pre-refactor tech-debt sweeps. Dispatches the auditor subagent for the broad sweep, optionally fans out to security-auditor / debugger on top findings (only when --depth=deep)…
workflow-hotfix
Branch-based P0 incident hotfix lifecycle — skips the worktree provider, opens a ready-for-review PR immediately after a minimal hotfix, ahead of verification and review (deferred by design), stamps a deferred-verification marker in the PR body, then reconciles once a regression test lands and the suite goes green.…