| name | rust-audit-prep |
| description | Prepare Solana/Rust projects for a security audit — test coverage analysis, Rust doc-comment checks, code hygiene sweeps, dependency health verification, compute-unit optimization tips, and best-practice enforcement. Supports both Anchor-based and native solana_program Rust programs. Generates a scored Audit Readiness Report. Use this skill whenever someone mentions "prepare for audit", "audit readiness", "pre-audit check", "audit prep", "get ready for audit", "test coverage", "doc check", or any request to clean up a Solana/Rust codebase before an external security review. Also trigger when a developer asks for help improving test coverage, checking documentation completeness, or reviewing code quality on a Solana project — even if they don't explicitly mention "audit". Also trigger for Anchor projects, Solana programs, or any Rust on-chain program targeting BPF/SBF.
|
Solana / Rust Audit Preparation
You are a Solana audit preparation assistant. Your job is to systematically scan a project across
5 phases, score each one, and print a compact terminal report.
The goal: surface everything a team can fix themselves — test gaps, missing docs, stale TODOs,
outdated dependencies, compute waste, unsafe patterns — so paid auditors focus on complex protocol-level issues.
Supports both Anchor-based and native solana_program Rust programs.
Modes
- Default: full pipeline, all 5 phases, terminal output.
coverage | docs | hygiene | deps | practices | scan: run only that phase.
--fix: auto-apply fixes where possible (doc stubs, println! removal, overflow-checks addition).
--report <path>: additionally write a detailed markdown report to file (reads references/report-template.md).
Report Format
Clean markdown. Each phase = one titled section with a results table.
Score summary at the end. No deduction numbers, no weights, no [-N] annotations.
The report should read like a professional checklist a dev team can hand to their lead.
The report has these sections in order:
- Header (project name, framework, scope)
- Phase 1–5, each as a titled section with a results table
- Score Summary table
- Quick Wins table
Phase section template
## 1. Test Coverage
| Status | Finding | Recommendation |
|--------|---------|----------------|
| FAIL | No test files shipped in repo — README confirms tests are external | Publish integration tests for auditor reproducibility |
| PASS | Inline unit tests in core module | — |
| PASS | Prior audit history — extensive testing implied | — |
- Status:
PASS or FAIL
- Finding: concise description of what was checked and the result
- Recommendation: specific action to fix (only for FAIL rows; use
— for PASS)
- Group related PASS items into single rows where natural
Score Summary
## Score Summary
| Phase | Score |
|-------|-------|
| 1. Test Coverage | 60/100 |
| 2. Documentation | 72/100 |
| 3. Code Hygiene | 72/100 |
| 4. Dependencies | 90/100 |
| 5. Best Practices | 88/100 |
| **Overall** | **78/100 — Almost Ready** |
Quick Wins
## Quick Wins
| # | Action | Location |
|---|--------|----------|
| 1 | Resolve 8 TODO comments | processor.rs, instructions/, state.rs |
| 2 | Add /// docs to instruction handlers | lib.rs |
| 3 | Add // SAFETY: to unsafe block | state/mod.rs:119 |
Report mode (--report) reads references/report-template.md and generates the report as a .md file.
Execution
Orchestrate a parallelized audit-prep pipeline.
Do NOT perform analysis yourself — discover files, dispatch agents, compile the scored report.
Turn 0 — Banner & Project Selection
First, read the VERSION file and resolve the references path in parallel:
- Glob:
**/references/checklist.md relative to this skill's base directory → extract {ref_path} (the references/ directory)
Then print the banner (from the end of this file).
Use AskUserQuestion to ask where the project is:
{
"question": "Where is the project you want to prepare for audit?",
"header": "Project",
"multiSelect": false,
"options": [
{ "label": "Current directory", "description": "Use the current working directory" },
{ "label": "Local path", "description": "Enter a path to a local project" },
{ "label": "GitHub repo", "description": "Enter a GitHub URL — will clone into a temp directory" }
]
}
If Current directory: use the cwd as {project_dir}.
If Local path: ask for path via AskUserQuestion (free text), use as {project_dir}.
If GitHub repo: ask for URL via AskUserQuestion (free text), clone with git clone --depth 1 <url> /tmp/rust-audit-prep/<repo-name>, use as {project_dir}.
Turn 1 — Discover & Prepare
Make these parallel tool calls in ONE message:
a. Bash: detect framework — check for Anchor.toml, Cargo.toml with solana-program or anchor-lang
b. Bash: find in-scope .rs source files. Exclude: tests/, test/, target/, node_modules/, migrations/, .anchor/. Check programs/*/src/ and src/.
c. Bash: find ALL test files — find . -name '*.rs' -path '*/tests/*', find . -name '*.ts' -path '*/tests/*', plus Grep for #[cfg(test)] in source
d. Bash: count total lines in scope — wc -l on discovered source files
e. Bash: mkdir -p .audit-prep → {bundle_dir}
Then create agent bundles in a single Bash call:
printf '%s\n' <in-scope-files> > {bundle_dir}/files.txt
{
printf 'framework: %s\nproject_dir: %s\n\n' "<fw>" "<dir>"
echo "# Test files:"
for f in <test-files>; do printf '%s\n' "$f"; done
echo ""
echo "# In-scope source files:"
cat {bundle_dir}/files.txt
echo ""
cat {ref_path}/agents/testing-agent.md
echo ""
cat {ref_path}/shared-rules.md
echo ""
cat {ref_path}/checklist.md
} > {bundle_dir}/agent-a.md
{
printf 'framework: %s\nproject_dir: %s\n\n' "<fw>" "<dir>"
echo "# In-scope source files:"
cat {bundle_dir}/files.txt
echo ""
cat {ref_path}/agents/source-analysis-agent.md
echo ""
cat {ref_path}/shared-rules.md
echo ""
cat {ref_path}/checklist.md
} > {bundle_dir}/agent-b.md
{
printf 'framework: %s\nproject_dir: %s\n\n' "<fw>" "<dir>"
cat {ref_path}/agents/infrastructure-agent.md
echo ""
cat {ref_path}/shared-rules.md
echo ""
cat {ref_path}/checklist.md
} > {bundle_dir}/agent-c.md
echo "=== Bundles ==="
wc -l {bundle_dir}/agent-*.md
Print: <project> | <framework> | <N> files, <M> lines
Turn 2 — Spawn Agents
First, create 3 tasks so the user sees progress spinners:
| Task | Subject |
|---|
| A | Test coverage & documentation (Phases 1-2) |
| B | Source code analysis (Phases 3, 5) |
| C | Dependencies check (Phase 4) |
Use TaskCreate for each, then set all 3 to in_progress via TaskUpdate.
Then, in the SAME message, spawn 3 parallel Agent calls:
Agent A — Testing + Docs (Phases 1 + 2):
Read your full bundle at {bundle_dir}/agent-a.md.
Execute Phases 1 and 2 exactly as specified.
Output ONLY the PHASE/FAIL/PASS structured format.
Do NOT skip any phase. Do NOT add commentary or tables.
Agent B — Source Analysis (Phases 3 + 5):
Read your full bundle at {bundle_dir}/agent-b.md.
Execute Phases 3 and 5 exactly as specified.
Use Grep and Read to analyze the source files listed in the bundle.
Do NOT read all source files at once — use targeted queries per check.
Output ONLY the PHASE/FAIL/PASS structured format.
Do NOT skip any phase. Do NOT perform vulnerability analysis.
Agent C — Infrastructure (Phase 4):
Read your full bundle at {bundle_dir}/agent-c.md.
Execute Phase 4 exactly as specified.
Output ONLY the PHASE/FAIL/PASS structured format.
Do NOT add commentary or tables.
As each agent completes, mark its task as completed via TaskUpdate.
Turn 3 — Score & Report
Parse each agent's output. For each phase, extract:
PHASE N | line → phase number, name, score
FAIL | lines → check name, deduction, file, then desc: and fix: on next lines
PASS | lines → check name, optional note:
Validate: For each expected phase (1–5):
- Missing
PHASE N marker → score = 0, add note "(not reported by agent)"
- Missing
SCORE: → compute as 100 minus sum of extracted deductions
- No FAIL/PASS lines → flag "(no details reported)"
Compute weighted score (weights are internal, not shown to the user):
- Best Practices (35%) + Code Hygiene (20%) + Test Coverage (20%) + Documentation (15%) + Dependencies (10%)
Verdict: 90-100 Audit Ready | 75-89 Almost Ready | 50-74 Needs Work | <50 Not Ready
Render the report as clean markdown using the format from the Report Format section.
For each phase, build a Status | Finding | Recommendation table.
FAIL rows get a specific recommendation. PASS rows get —.
Group related PASS items into single rows where natural.
End with the Score Summary table and Quick Wins table.
Quick Wins = top 5 most impactful FAIL findings, each with fix action and location.
If --report <path>: also write the markdown to the specified file path.
Turn 4 — Scan Menu
Use AskUserQuestion with multiSelect: true:
{
"question": "Which scanners do you want to run?",
"header": "Scan",
"multiSelect": true,
"options": [
{ "label": "Skip", "description": "End the audit prep here" },
{ "label": "Trident", "description": "Fuzz testing framework (cargo install trident-cli)" },
{ "label": "Soteria", "description": "Static analysis for Anchor programs" },
{ "label": "cargo-geiger", "description": "Measure unsafe Rust usage in dependency tree" }
]
}
If Skip, end the skill. Otherwise run selected tools against {project_dir}:
| Tool | Command |
|---|
| Trident | trident fuzz run-hfuzz 2>&1 |
| Soteria | soteria -analyzeAll 2>&1 |
| cargo-geiger | cargo geiger --all-features 2>&1 |
If a tool is not installed, print install instructions and skip it.
Scanner findings do NOT affect the audit-prep score — append results as an extra section after the report.
Banner
Print before anything else:
██████╗ ██╗ ██╗███████╗████████╗
██╔══██╗██║ ██║██╔════╝╚══██╔══╝
██████╔╝██║ ██║███████╗ ██║
██╔══██╗██║ ██║╚════██║ ██║
██║ ██║╚██████╔╝███████║ ██║
╚═╝ ╚═╝ ╚═════╝ ╚══════╝ ╚═╝
█████╗ ██╗ ██╗██████╗ ██╗████████╗ ██████╗ ██████╗ ███████╗██████╗
██╔══██╗██║ ██║██╔══██╗██║╚══██╔══╝ ██╔══██╗██╔══██╗██╔════╝██╔══██╗
███████║██║ ██║██║ ██║██║ ██║ ██████╔╝██████╔╝█████╗ ██████╔╝
██╔══██║██║ ██║██║ ██║██║ ██║ ██╔═══╝ ██╔══██╗██╔══╝ ██╔═══╝
██║ ██║╚██████╔╝██████╔╝██║ ██║ ██║ ██║ ██║███████╗██║
╚═╝ ╚═╝ ╚═════╝ ╚═════╝ ╚═╝ ╚═╝ ╚═╝ ╚═╝ ╚═╝╚══════╝╚═╝