Project Agentification (experimental)

Assess, harden, scaffold, and diagnose a repository's agent-readiness. Harness-agnostic; portable-first (AGENTS.md, SKILL.md, MCP, OpenTelemetry). Grounding sources live in skill.json; this file is runtime routing only.

Core principle

Make the repository legible to a stateless, context-window-bound LLM that must rediscover the project on every turn — through deterministic structure, hand-curated instructions (never autogenerated), executable boundaries, and replayable feedback loops. Three load-bearing claims surface in every playbook:

1. Don't /init. Hand-curate from observed failures.
2. Token budget is the dominant scarcity. Prefer on-demand → on-trigger → always-loaded.
3. Hard gates over soft prose. Hooks at 100% vs prose at 70%.

Scope

For any repo that wants coding agents (Claude Code, Cursor, Codex, Copilot, Aider, Windsurf, AGENTS.md-compatible harnesses) to work well in it. This skill makes no assumptions about eval infrastructure, benchmarks, telemetry, or run-level measurement — it works from project knowledge (stack, layout, monorepo scope, build/test commands, top-level invariants).

If you also have a feedback signal — eval suites, run-level telemetry, A/B baselines, or a skill catalog under test — pair this skill with evidence-driven-agent-rules, which layers a reflection-log workflow on top (capture observed agent failures, promote recurring patterns into AGENTS.md rules via the W1 ≥3-entries floor). This skill does not require it; most repos won't.