ottersec-prep— agent skill

Preamble (run first)

# Suiperpower telemetry. Routes through the CLI so projects.json + Convex stay in sync.
# Silent on failure so the skill never blocks. Opt out: SUIPERPOWER_TELEMETRY=off.
#
# AGENT NOTE: when this skill finishes, run the matching completion command:
#   suiperpower track ottersec-prep build completed
# Or use "failed" / "aborted" if it ended that way.
command -v suiperpower >/dev/null 2>&1 && suiperpower track ottersec-prep build started >/dev/null 2>&1 &
true

If TEL_PROMPTED is no, before doing real work, ask the user:

Help suiperpower get better. We track which skills get used and how long they take. No code, no file paths, no PII. Change anytime in ~/.suiperpower/config.json.

>

A) Sure, anonymous B) No thanks

Write the answer to ~/.suiperpower/config.json telemetryTier field and create ~/.suiperpower/.telemetry-prompted. Then continue.

What this skill does

Walks the OtterSec pre-audit checklist on the user's Move package, surfaces the gaps, and produces the engagement package the auditor expects (threat model, architecture diagram outline, test report, known-issues doc). The goal is for audit time to be spent on real findings, not on hygiene noise.

When to use it

• About to engage OtterSec or another Sui Move auditor.
• Wanting to self-assess audit readiness.
• Preparing the audit-prep portion of a Sui Overflow submission targeting the OtterSec track.

When NOT to use it

• Pre-MVP code where auditing is premature; use review-move for an in-house P0-P3 walk first.
• If the project does not have any Move code yet, use build-with-move first.
• For non-Move security questions (frontend XSS, key custody for sponsor wallets), use the relevant build skill.

If you activated this and the user actually wants something else, consult skills/SKILL_ROUTER.md and hand off.

Inputs

• A Sui project with at least one Move package under move/ or similar.
• Optional: .suiperpower/build-context.md and any prior review-move output.

If unclear, interview the user for:

• Which packages are in scope for the audit?
• Target deploy: which network, what value at risk?
• Are there parts the user explicitly does not want audited (test scaffolding, prototypes)?
• What is the timeline for the audit and the deploy?

Outputs

• A filled-in pre-audit checklist (every P0 item answered, P1-P3 documented).
• A THREAT_MODEL.md in the project: who is trusted, what are the assets, what are the known risks.
• An architecture summary (which modules talk to which, where capabilities flow).
• A test-coverage report from sui move test.
• A KNOWN_ISSUES.md listing decisions to ship despite known concerns.
• Append to .suiperpower/build-context.md:

  ## ottersec-prep session, <timestamp>
  - packages in scope: <list>
  - checklist completion: <P0 percent>, <P1 percent>
  - threat model doc: <path>
  - test count: <n>, coverage: <percent>
  - open issues: <list>

The skill never deletes files outside the integration source path without explicit user confirmation.

Workflow

1. Context gathering

• Read .suiperpower/build-context.md if present.
• Confirm the package(s) in scope and the deploy target.

1. P0 walk

• Step through every P0 item from skills/data/guides/security-checklist.md.
• For each, mark pass / fail / not-applicable with a one-line note.
• For fails, propose a fix and let the user decide whether to fix now or document as known.

1. Capability inventory

• List every *Cap struct in the package(s).
• For each, document: who creates it, who holds it, who can pass it to whom, whether it leaks via Display or public read.

1. Visibility audit

• Confirm public(package) is the default; public is reserved for true cross-package APIs.
• Flag any leftover friend declarations; recommend migration.
• Confirm #[test_only] on test helpers.

1. Init function audit

• Confirm one-time witness pattern intact.
• Confirm init does not leak capabilities outside its intended distribution.
• Confirm reinitialization is impossible (no shared init paths that can be re-entered).

1. Test coverage

• Run sui move test, capture output.
• Confirm every public entry function has at least one positive and one expected-failure test (where applicable).

1. Build cleanliness

• Run sui move build. Confirm zero warnings.
• Strip assert!(false), commented-out checks, and std::debug::print from production paths.
• Confirm error codes are named constants, not magic numbers.

1. Dependency pin

• Confirm Move.toml uses pinned revs / tags. Reject main or floating versions in production deps.

1. Threat model + known issues

• Help the user write THREAT_MODEL.md and KNOWN_ISSUES.md if absent.
• Surface assumptions the auditor needs to know.

1. Engagement package

• Compose the audit submission: repo URL, scope, threat model, architecture, test report, checklist, known issues.
• Surface the OtterSec contact path (https://osec.io/).

1. Writeback

• Append session details to .suiperpower/build-context.md.

Quality gate (anti-slop)

Before reporting done, the skill asks itself the following and refuses to declare success if any answer is no:

• Is every P0 item from the security checklist explicitly marked pass / fail / N/A with a note?
• Has every public entry function been mapped to at least one happy-path test and one expected-failure test where applicable?
• Is THREAT_MODEL.md non-trivial (assets, trust assumptions, known risks named)?
• Is the capability inventory complete (every *Cap listed with holder strategy)?
• Is Move.toml pinned, no main in production deps?
• Is the engagement package something the user could send to OtterSec today without a follow-up "actually, I forgot..."?

If any answer is no, the skill reports the gap and works through it before claiming the package is audit-ready.

References

On-demand references (load when relevant to the user's question):

• references/pre-audit-walkthrough.md: Item-by-item walk of the checklist with example pass/fail patterns.
• references/finding-categories.md: Common findings OtterSec produces, with mitigation patterns.
• references/threat-model-template.md: One-page threat-model template.

Knowledge docs (load when scope expands beyond what is in references):

• skills/data/sui-knowledge/sponsor-docs/ottersec-checklist.md: Full checklist source.
• skills/data/guides/security-checklist.md: P0-P3 reference.

Use in your agent

• Claude Code: claude "/suiper:ottersec-prep <your message>"
• Codex: codex "/ottersec-prep <your message>"
• Grok Build: run grok, then /ottersec-prep <your message> in the session
• Cursor: paste a chat message that includes a phrase like "prepare for OtterSec audit", or load ~/.cursor/rules/ottersec-prep.mdc and reference it.

If you activated this and the user actually wants something else, consult skills/SKILL_ROUTER.md and hand off.