DevFlow Roadmap: Post-v1.0 Enhancement Plan

[dean0x]

DevFlow Roadmap: Post-v1.0 Enhancement Plan

North Star tracking issue for all post-v1.0 enhancements. Sourced from competitive analyses (#107, #114) and internal feature planning.

---

What We're Building

Priority	Issue	Feature	Effort	Wave
1	#113	Reviewer Confidence Thresholds (>80% filter)	Small	1
2	#91	Version Manifest for Upgrades	Small	1
3	#111	Search-First Skill	Small	1
4	#99	Persistent Project Knowledge (ADRs, patterns, pitfalls)	Medium	2
5	#108	/gap-analysis + /plan Commands	Large	3
6	#115	Continuous Learning / Instinct System	Large	4
7	#98	PreToolUse Hook Enforcement	Medium	5
8	#110	PostToolUse Quality Hooks (format, typecheck)	Medium	5
9	#112	Stack-Specific CLAUDE.md Templates	Medium	5
10	#116	Model Profiles (quality/balanced/budget)	Medium	5
11	#117	Health Check Command (/health)	Small	5

---

What We're Deferring

Issue	Feature	Effort	Why Defer
#27	Shepherd Hooks	Large	Hooks API still evolving
#100	Multi-Perspective Planning	Small-Med	Uncertain ROI, adds tokens
#101	Elicitation Techniques for /specify	Small	/specify works fine as-is
#109	Context Modes (dev/research/review)	Small	Third dimension, marginal gain
#103	Dispatch Gating via File Markers	Small	Depends on #98; problem not yet hit
#92	Rollback on Partial Install	Medium	Install is idempotent; failures rare
#118	Decision Preservation (Locked/Flexible)	Medium	Adds ceremony to /specify flow
#119	Checkpoint-Driven Workflows	Medium	Nice but complex plumbing
#120	De-Sloppify Simplifier Categories	Small	Do when we touch Simplifier next
#121	Stub Detection for Scrutinizer	Small	Scrutinizer already catches most
#122	Wave-Based Parallel Execution	Medium	Complex; validate need first
#123	/quick Command	Medium	Validate the gap exists first
#124	Goal-Backward Verification	Medium	Defer with #27

---

Waiting on Decision

Items we're not acting on yet. May revisit.

Issue	Feature	Reason for Hold
#107	Harness Alpha Research Report	Informational — all ideas tracked individually now
#114	Anvil Research Report	Same — all ideas tracked individually
#89	Settings.json Concurrent Write Protection	Theoretical race; never reported
#90	Skill Namespace Prefix	Premature; no ecosystem conflict exists
#102	Lore Engine / Narrative UX	Cosmetic; zero impact on quality
#6	Symlink Attack Prevention	Requires pre-positioned local attacker
#7	Security Warnings During Install	Marginal value for a dev tool
#8	Script Integrity Verification	Post-install tampering needs compromised FS
#9	Cryptographic Script Signing	Massive effort, advanced threat model
#10	Sandboxed Script Execution	Research-only, theoretical risk
#125	Multi-IDE Adapter Layer	Premature
#126	Session Aliasing/Management	CLI sugar, not a differentiator
#127	Cost Tracking	Claude Code itself may add this
#128	Package Manager Detection	Windows-specific edge case
#129	Bridge Files for Inter-Hook Comms	Pattern, not a feature — use when needed
#130	Eval-Driven Development Metrics	Academic; unclear practical value
#131	Hook Profile Gating	Premature without richer hook ecosystem
#132	Iterative Retrieval for Subagents	Over-engineering current model
#133	Persistent Codemaps	High maintenance burden
#134	Security Audit Command	Overkill for current user base
#135	Per-Task Atomic Commits	Too granular; noisy git history
#136	Session Handoff Files	WORKING-MEMORY.md already does this

---

Keeping As-Is

Issue	Feature	Status
#23	Tech Debt Backlog	Living document — stays open, not a standalone task

---

Wave Plan

Wave	Issues	Theme	Parallelizable
1	#113, #91, #111	Quick wins — confidence filter, version tracking, search-first	Yes, all 3 independent
2	#99	Persistent knowledge — foundation for learning	Solo
3	#108	/gap-analysis + /plan — biggest workflow feature	Solo (large)
4	#115	Continuous learning — builds on #99	Solo (large)
5	#98, #110, #112, #116, #117	Hook enforcement + templates + utilities	Partially parallel

[dean0x]

Roadmap update for v1.8.0 agent quality work

Moved to completed (via #154)

• feat: De-Sloppify Categories for Simplifier Agent #120 De-Sloppify Simplifier Categories → Simplifier now has 8 structured slop detection categories
• feat: Stub Detection Patterns for Scrutinizer #121 Stub Detection for Scrutinizer → New stub-detection.md reference + responsibility in Scrutinizer
• feat: Goal-Backward Verification for Shepherd #124 Goal-Backward Verification → Shepherd now has goal-backward verification, artifact depth, stub type, re-verification

Closed

• feat: Shepherd hooks for subagent orchestration #27 Shepherd Hooks → useful parts extracted into feat: Goal-Backward Verification for Shepherd #124; hooks infrastructure deferred indefinitely

Wave reorder

• Wave 3: feat: Continuous Learning / Instinct System #115 Continuous Learning / Instinct System (was Wave 4) — foundation now laid by feat: De-Sloppify Categories for Simplifier Agent #120/feat: Stub Detection Patterns for Scrutinizer #121/feat: Goal-Backward Verification for Shepherd #124
• Wave 4: feat: Iterative Retrieval for Subagents #132 Iterative Retrieval for Subagents (was deferred/Waiting on Decision)
• Wave 5+: feat: add /gap-analysis and /plan commands with /implement plan detection #108 /gap-analysis + /plan (was Wave 3) — drops lower in priority