Collection · sigil/launch-set · v1.0.0 · MIT
Sigil Launch Set
The first seven Sigil skills, each with its own proof suite, for the APIs and safety rules coding agents most often get wrong in late 2026.
Seven skills that carry current, precise knowledge a coding agent otherwise gets wrong: GitHub Actions hardening, zero-downtime PostgreSQL migrations, MCP servers on the v2 TypeScript SDK, Vercel AI SDK 7 tools and structured output, Express 5, Zod 4 and accessible React widgets. Every member ships with a three-task suite whose hidden verifier fails on the starting repo and passes on the reference solution, so its effect can be measured rather than assumed. Installing the bundle installs each member in its own native format; each member keeps its own digest and evidence.
- Backend
- DevOps
- AI engineering
- Frontend
- build
- secure
- ship
- maintain
- typescript
- nodejs
- github-actions
- npm
- postgresql
- mcp
- ai-sdk
- express
- zod
- react
uplift
No runs exist for this digest.
Not measured for this digest.
Untested is not the same as useless: nobody has measured it yet. Proof comes from A/B runs on hidden tests:
npx sigil prove catalog/collections/sigil/launch-set --suite <suite> --runs 6trust level · submitted
- in the catalog: met
- structure check: not met
- static scan, no high findings: not met
- canary sandbox passed: not met
- maintainer approved: not met
canary sandbox
Not run for this digest.
static scan
Not run for this digest.
compatibility · targeted, compiled and evaluated are separate claims
| agent | targeted by author | compiled | evaluated in the lab | notes |
|---|---|---|---|---|
| Claude Code | targeted: yes | no record for compiled | no record for evaluated | |
| Codex | targeted: yes | no record for compiled | no record for evaluated | |
| Cursor | targeted: yes | no record for compiled | no record for evaluated | |
| GitHub Copilot | targeted: yes | no record for compiled | no record for evaluated | |
| Gemini CLI | targeted: yes | no record for compiled | no record for evaluated | |
| OpenCode | targeted: yes | no record for compiled | no record for evaluated |
declared permissions
- network
- github.com, api.github.com
- shell
- node, npm test, npm ls, npm run typecheck, npx tsc, grep, rg, psql, git ls-remote, gh api, zizmor, actionlint
- filesystem
- write
- env
- none
Undeclared access is a finding in the scan and a tripwire in the canary.
source
- path
- catalog/collections/sigil/launch-set
- authors
- dgesteves
- entry
- BUNDLE.md (bundle)
- registry
- 2026-11-02 · 5cb851d
install
$ npx sigil add sigil/launch-set --agent claude-codeNo compile record for Claude Code on this digest. Install never runs capability scripts, and refuses on any digest mismatch.
known limitations
- Results for the bundle measure all seven skills installed together. A member can help on its own suite and still be diluted or crowded out when every skill is loaded; read each member's own result too.
- Every member is pinned to the versions it was checked against (November 2026). Projects on older majors (AI SDK 5-6, MCP SDK v1, Express 4, Zod 3, PostgreSQL before 12) get migration notes, not full coverage.
- The suites are small (three tasks each), so early results are likely to be inconclusive until more runs and tasks exist.