Skip to content

ROADMAP.md

Latest commit

 

History

History
44 lines (39 loc) · 6.46 KB

ROADMAP.md

File metadata and controls

44 lines (39 loc) · 6.46 KB
Raw

Devkit Roadmap

Implemented

  • MCP engine — Go server exposes devkit_start, devkit_advance, devkit_status, devkit_list tools inside Claude Code. Step ordering enforced via MCP tool scoping + PreToolUse hook exit 2. Session state in session.json (hot path, <50ms hook reads) + SQLite (cold history). ~65% token reduction vs old monolithic prompts.
  • Skills-first architecture — All entry points are skills in skills/ (tab-completable slash commands in current Claude Code; bare names like /tri-review work, /devkit:<name> form also works for disambiguation). Primary user-facing commands: /tri-review, /tri-debug, /tri-security, /health, /setup-rules (user-only via disable-model-invocation). Every workflow also has a dedicated skill for natural-language dispatch. The commands/ directory is retained for backward compat but empty of new entries — redundant tri-* command files were removed (skills take precedence per Claude Code docs) and the generic /devkit:workflow <name> runner was removed since every workflow now has its own slash command (/feature, /bugfix, etc.).
  • Deterministic workflow conversion — All command logic moved from LLM-interpreted markdown to Go-engine-driven YAML workflows; ~3,600 lines of inline logic removed
  • 39 skills — 22 workflow trigger skills (feature, bugfix, refactor, audit, harness-audit, research, deep-research, pr-ready, autoloop, test-gen, doc-gen, onboard, tri-review, tri-debug, tri-security, tri-dispatch, self-audit, self-improve, self-lint, self-migrate, self-perf, self-test) + 7 coding principles (executing, clean-code, DRY, YAGNI, dont-reinvent, stuck, scratchpad) + 4 tools (gcli, scrape, screenshot, browser) + 1 meta-orchestration (mega-pr) + 2 content (changelog, adr) + 1 reference (creating-workflows)
  • Deterministic skill dispatch for every workflow — Every one of the 22 workflows has a natural-language trigger skill with keyword-rich description. Saying "build a feature", "tri review", "deep research X", etc. deterministically invokes the matching skill, which calls devkit_start and the engine enforces every step from there. Closes the entry-gate non-determinism where 11/18 workflows previously had no natural-language path. Skill tool added to the guard allowlist so nested mid-workflow skill dispatch works.
  • 6 agents — Scoped tool access, worktree isolation, model assignment
  • 12 hooks — Safety (destructive command blocking, edit-time security patterns, PR gate), observability (audit trail, slop detection, post-validation, subagent verification, language-aware code review), optimization (RTK token compression), workflow enforcement (devkit-guard, devkit-stop-guard)
  • Graceful degradation — tri:* commands work with 1-3 agents depending on installed CLIs
  • Goal decomposition — Task DAG with dependency ordering and parallel execution
  • Concurrency limits — Max 3 parallel agents in multi-agent commands
  • Early-exit conditions — Self-improvement loops stop when goal is met, not just at max iterations
  • Token budget guidance — Per-command budget recommendations with model downgrade patterns
  • RTK token optimization — Optional PreToolUse hook compresses Bash output via RTK (60-90% savings)
  • 22 YAML workflows — Portable workflow definitions (feature, bugfix, refactor, research, deep-research, autoloop, harness-audit, self-, tri-, test-gen, doc-gen, onboard)
  • Separate marketplace — Multi-plugin marketplace at 5uck1ess/marketplace
  • Companion ecosystem — Evaluated official marketplace, documented holistic setup with 7 complementary plugins
  • Hypothesis-driven perf — Evidence gathering, ranked hypotheses, one-at-a-time testing replaces blind benchmark loops
  • Post-PR monitoring — CI watching + iterative reviewer comment resolution
  • AST repo mapping — Symbol index with dependency graph, cached for agent navigation
  • Generic YAML workflow engine — Deterministic step execution, branching, loops, parallel dispatch, budget enforcement in compiled Go
  • Triage-based phase skipping — TINY/SMALL/MEDIUM/LARGE classification with fast paths
  • Iteration scratchpads — Persistent memory across loop iterations to prevent repeated failures
  • Cross-domain dirty-bit enforcement — Blocks completion without test evidence per domain
  • Language-universal hooks — Consolidated language-specific hooks into lang-review.sh with Go, TypeScript, Rust, Python, and Shell support
  • Hook consolidation — Merged 14 hooks into 10, reduced per-edit shell processes from 7 to 4
  • Local runner — OpenAI-compatible HTTP runner for Ollama/llama-server/llama-swap/vLLM/SGLang/LM Studio/LocalAI alongside Claude/Codex/Gemini CLI runners. Opt-in via DEVKIT_LOCAL_ENABLED=1; defaults to qwen3:32b at localhost:11434/v1. Intended for fast-tier dispatch; cloud runners stay default for smart/general given local tool-call reliability gap. devkit-engine probe-local verifies endpoint reachability, latency, and configured-model presence with human or --json output (wired into /devkit:health).
  • Stealth scraping backends — scrape skill adds Camoufox (patched Firefox with leak-fixed JA3/TLS fingerprints) between Playwright and Firecrawl with auto-fallback on Cloudflare/DataDome/Incapsula challenge-page signatures, plus Scweet as host-routed backend for x.com/twitter.com URLs when SCWEET_AUTH_TOKEN is set.

Retired

Items below were on the roadmap but determined to be unnecessary — either already solved by existing infrastructure or too speculative to justify the complexity.

Item Why removed
Stop hook redesign Still fires every turn, but exits early with approve when no files are changed — near-instant on clean trees, so the performance concern is moot. Revisit only if it causes measurable latency.
Cost event hooks Budget enforcement already exists in the Go engine via overBudget() + addCost() callbacks with hard limits
Execution registry Step tracking already handled by SQLite via lib.DB with status, cost, and timing per step
Preset library The 18 YAML workflows and 22 skills already serve this purpose
Framework-specific review checklists lang-review.sh covers language-level patterns; framework-specific rules are better added per-project via hookify
Conditional hook firing Hooks already self-filter internally (extension checks, changed-file checks); a generic condition system adds complexity for no current need