CtrlK
BlogDocsLog inGet started
Tessl Logo

iikit

Intent Integrity Kit: spec-driven development pipeline (constitution->specify->plan->checklist->testify->tasks->analyze->implement). Consumes upstream intent-integrity-chain/kit conventions. Topics: 00-constitution, 01-specify, 02-plan, 03-checklist, 04-testify, 05-tasks, 06-analyze, 07-implement, 08-taskstoissues, bugfix, clarify, core.

SKILL.md
Quality
Evals
Security

iikit

Router skill for the Intent Integrity Kit: a spec-driven pipeline that turns intent into shipped code through ordered gates. [DOC]

When to use

Trigger on any spec-driven request: a constitution, spec, plan, checklist, tests, task list, or driving implementation/bugfix. Named stage (or its number 00–08) → that topic; described intent → infer the earliest unmet stage. Don't use for ad-hoc coding with no spec artifact in play. [INFERENCE]

Routing

  1. Resolve topic (enum is authoritative; numbers map 1:1). [CONFIG]
  2. Read EXACTLY ONE playbook from routes:. Never load the cluster — that defeats the router and burns context. [INFERENCE]
  3. Apply by depth: deep → exhaustive, verify each step; quick → essentials, single pass. [CONFIG]

Pipeline (each output feeds the next): 00-constitution → 01-specify → 02-plan → 03-checklist → 04-testify → 05-tasks → 06-analyze → 07-implement → 08-taskstoissues. clarify, bugfix, core are off-spine helpers. [DOC]

Inputs / Outputs

  • In: topic (required), depth (default quick), plus the upstream artifact the stage consumes (e.g. 02-plan needs a 01-specify spec). [CONFIG]
  • Out: the single stage artifact, evidence-tagged with the Alfa core set ([CODE] [CONFIG] [DOC] [INFERENCE] [ASSUMPTION]). [DOC]

Validation gate (before declaring done)

  • Exactly one playbook read; topic ∈ enum; predecessor artifact existed or was explicitly created (no stage on a missing predecessor). [INFERENCE]
  • Constitution v6.0.0 enforcement honored; script-first rule applied; tags from ONE family only. [CONFIG]
  • Score against assets/quality-rubric.json (QR-01…QR-08); see the assets/ bundle for the routing checklist. [CONFIG]

Anti-patterns & self-correction

  • Loading multiple playbooks "to be safe" — read one, re-route if wrong. [INFERENCE]
  • Guessing topic when genuinely ambiguous → ask, don't fan out. [ASSUMPTION]
  • Skipping a predecessor stage to reach implementation faster. [INFERENCE]
  • If the chosen playbook doesn't fit mid-execution: stop, re-resolve topic, read the correct one — never stack a second on top. [INFERENCE]

Spine: Discover → Analyze → Execute → Validate. [DOC]

Provenance — tessl tile dependency [DOC]

iikit is a thin overlay over the tessl tile tessl-labs/intent-integrity-kit. Its deeper assets — iikit-core (references like constitution-loading, formatting-guide, phase-separation-rules, model-recommendations; templates like spec/plan/tasks/checklist) and the helper scripts under .tessl/tiles/.../iikit-core/scripts/ — are provisioned by tessl at install time, NOT vendored in this repo. [CONFIG]

  • When a playbook names one of those assets in backticks (e.g. formatting-guide, spec-template), read it from the installed tile, not from skills/iikit/. [DOC]
  • If tessl is not installed (.tessl/ absent), those assets are unavailable — run the spine on the in-repo playbooks (references/00-08, bugfix, clarify, core) and flag the missing tile. [ASSUMPTION]
  • Do NOT fabricate local copies of iikit-core content — it is maintained upstream; duplicating it would drift. [DOC]
Repository
JaviMontano/jm-adk-beta
Last updated
First committed

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.