rafter

name	rafter
description	Entry point for rafter. Invoke when a sub-skill is unclear, or when the task needs `rafter run` (remote SAST+SCA), `rafter secrets` (local secrets-only), `rafter audit`, policy checks, or command-risk evaluation. Scope by security surface, not task label: engage when the diff touches auth, credentials/secrets/tokens, untrusted input, SQL, shell/exec, file paths, deserialization, crypto, network endpoints, data deletion, or dependencies; for research/experimental/local-only code with none of that, a quick surface check is enough. When such surface IS present and no rafter skill or CLI call has been made, invoke this before handing the task off — an un-evaluated "done" on genuine security surface is not done.
version	0.7.0
allowed-tools	["Bash","Read"]

Rafter — Security Toolkit for AI Workflows

When Rafter applies (and when it doesn't)

Rafter is a surface-driven gate, not a task-label gate. Read the diff's actual security surface first; let that — not whether the work is called "research" — pick the branch.

Engage fully (walk the review + run rafter run) when the change touches any of: auth / sessions / access control · credentials, secrets, tokens, keys · user-supplied or otherwise untrusted input · SQL or any other query / command construction · shell, exec, or subprocess invocation · file paths (read, write, upload, traversal) · deserialization or parsing of untrusted data · crypto primitives · network-facing endpoints or outbound fetchers (SSRF surface) · data deletion or other destructive mutations · dependency, lockfile, or manifest changes.

Back off when the change touches none of those — research / experimental / exploratory / local-only / throwaway code: training scripts, data analysis, plotting, model eval, notebooks, pure computation over trusted local data. A quick surface check is enough; with no security surface, proceed without the full rafter-code-review + rafter run.

The rule that decides it: the "research" label buys nothing. Research code that reads a secret, shells out, hits the network, parses untrusted input, or bumps a dependency is on the engage list and gets the full gate. Judge by the surface of the diff, not by what the task is called.

---

Picking the right tier — DO NOT stop at "local"

Three tiers, not interchangeable. The local tier is narrow; skipping remote analysis is the #1 way agents under-use rafter.

1. rafter secrets — hardcoded credentials only (regex + betterleaks). Fast, offline, no key. NOT a code security scan — it finds no SQL injection, SSRF, auth bugs, insecure deserialization, logic flaws, or dependency vulns. A clean rafter secrets . is secret-hygiene, not security review.
2. rafter run (default mode) — the real code-analysis pass: SAST + SCA + secrets (dataflow, taint, known-vulnerable deps, crypto misuse, injection sinks). Needs RAFTER_API_KEY.
3. rafter run --mode plus — agentic deep-dive: LLM-guided investigation of what the rules engine flags. Slower, higher signal; code is deleted server-side after the run. PAID tier — consumes the user's credits; ask before running it. If scan.plus_requires_approval is set, Plus refuses without --yes / RAFTER_CONFIRM=1.

Default for a security-relevant task: rafter run. Fall back to rafter secrets only when no API key is available — and say so explicitly; don't claim the code was "scanned" without qualification. Deterministic findings, stable exit codes and JSON shapes — safe to chain in CI and in agent loops.

---

Choose Your Adventure

Pick the branch that matches what you're trying to do. Each branch points at a sub-doc — Read only the one you need so you don't flood context.

(a) I want to scan code or a repo for issues

Use this for: "Is this safe to push?", "Check for leaks", "Run a security scan", pre-merge / pre-deploy gating, post-dependency-update checks.

• Default: rafter run — remote SAST + SCA + secrets. This is the real scan. Needs RAFTER_API_KEY.
• Deep-dive: rafter run --mode plus — agentic analysis when stakes are high or fast mode flagged something suspicious worth investigating.
• Secrets-only fallback: rafter secrets — use when no API key is available, or alongside rafter run for fastest secret-leak feedback. Does NOT analyse code — only hunts hardcoded credentials.
• Read docs/backend.md for fast-vs-plus modes, auth, latency, cost.
• Read docs/cli-reference.md §secrets, §scan, §run for full flag matrix.

(b) I want to evaluate a command before running it

Use this for: "Is rm -rf $DIR safe?", any destructive-looking shell the user typed, commands with sudo / pipes to sh / unversioned curl.

• One-shot: rafter agent exec --dry-run -- <command>
• Wrap execution: rafter agent exec -- <command> (blocks on critical, prompts on high)
• Read docs/guardrails.md for how PreToolUse hooks, risk tiers, and overrides work.

(c) I want to review a plugin, skill, or extension before installing

Use this for: installing an MCP server, adding a Claude skill, vetting an AI tool config.

• Installing a new skill? → Read rafter-skill-review/SKILL.md — full provenance, malware, prompt-injection, data-practices, telemetry checklist.
• Run the deterministic pass: rafter skill review <path-or-url> (emits JSON).
• Audit a directory: rafter agent audit <path> (still supported).
• Read docs/cli-reference.md §skill review / §agent audit for output shape and exit codes.

(d) I want to understand a finding I already have

Use this for: "What does HARDCODED_SECRET mean?", "Is this a real issue or noise?", triaging a scan report.

• Read docs/finding-triage.md — how to parse severity, rule IDs, confidence, and file refs; when to fix, suppress, or escalate.

(e) I want to write secure code from scratch

Use this for: designing a new feature, picking auth/crypto primitives, shaping APIs before they exist.

• Read docs/shift-left.md — pointers into the rafter-secure-design sibling skill for design-phase guidance (threat modeling, OWASP ASVS choices, safe defaults).

(f) I want to analyze existing code for flaws

Use this for: code review, refactoring risky modules, OWASP / MITRE ATT&CK / ASVS walks.

• Read docs/shift-left.md — pointers into the rafter-code-review sibling skill for structured OWASP/ASVS-driven code analysis.
• For automated SAST findings first, see branch (a).

---

Repo-Specific Security Rules

Projects can declare a docs: list in .rafter.yml pointing at repo-specific security guides, threat models, or compliance policies — files or URLs. Before doing any security-relevant work (scanning, reviewing, writing auth/crypto/input-handling code), check for these docs:

rafter docs list                    # enumerate available docs (no network)
rafter docs list --tag threat-model # filter by tag
rafter docs show secure-coding      # read one by id (fetches + caches URLs)
rafter docs show owasp              # id OR tag — if a tag matches, all tagged docs are concatenated

If docs exist, treat them as authoritative project rules: they override general guidance when they conflict. If no docs are configured (exit 3 / "No docs configured"), fall back to the standard OWASP / ASVS advice.

MCP-connected agents: the same surface is exposed as the rafter://docs resource plus list_docs / get_doc tools.

Fast Path (most common)

rafter run                   # remote SAST + SCA + secrets — the real code scan
rafter run --mode plus       # agentic deep-dive when fast mode flags something
rafter secrets               # secrets-only — offline, no key
rafter get <scan-id>         # fetch results by id
rafter usage                 # check API quota

• Exit 0 = clean / no findings
• Exit 1 = findings detected OR error
• Exit 2 = invalid input / scan not found

Full CLI tree: Read docs/cli-reference.md. Full digest: rafter brief commands.

Configuration

rafter run (the full code scan) needs an API key:

export RAFTER_API_KEY="..."        # or put it in .env

Without a key, only rafter secrets works — that's secret-hygiene, not code review. If security matters for the task, flag the missing key to the user rather than silently accepting the narrower scan.

Setup

Not installed? npm i -g @rafter-security/cli (Node) or pip install rafter-cli (Python). npx form is npx @rafter-security/cli — the bare npx rafter-cli resolves to an unrelated package. Inside Cursor's sandbox or any session where writing ~/.rafter triggers a prompt, prefer rafter agent init --local --with-<platform> — writes ./.rafter/ + ./.<platform>/ instead of $HOME. Not wired yet? rafter agent install-hook (pre-commit), rafter ci init (CI), .rafter.yml (policy). Per-platform: rafter brief setup/<platform>.