2026-06-30 · SpecKit Companion Part 2 demand-radar run #1

What the Community Wants

Part 2 — demand signals from GitHub issues, YouTube comments & ~35 articles. The audience side of the video trends report.

The one thing to remember: the loudest, most-repeated request across every source is "support my tool" — and that's exactly what SpecKit Companion does better than anyone. The field is fragmenting into a dozen single-ecosystem frameworks, and users are begging each one to work with their editor, CLI, and model. Companion doesn't pick a side; it orchestrates across them.

The issue trackers make it concrete: integration requests are the #1 unmet need in the category. Cross-provider isn't a nice-to-have — it's the wedge.

Four lanes: GitHub issues (Spec Kit, OpenSpec, Superpowers, BMAD — literal feature requests), YouTube comments, articles (10 you flagged + ~25 discovered), and Reddit (wired for the weekly run). This is also the first run of the new demand-radar skill — the baseline the weekly tracker builds on.

Demand themes (goal-scored)
🔥 hot/dominant ▲ rising ▬ steady goal served

D1 · "Support my tool" — cross-provider breadth

🔥 #1adoption + competitive

Where: every rival's top issues — Spec Kit Support Cursor CLI / Warp / Antigravity / Codex; BMAD OpenCode (41 comments), Augment (31); OpenSpec Zed, .agents installer; Superpowers Antigravity / Copilot / Gemini / codex / pi. Plus the whole Sonnet-5-vs-Opus comment war.

Move: lead with it everywhere. Competitors bolt on integrations one PR at a time; Companion orchestrates across providers by design. Each provider you add answers a request going unmet on a rival's tracker.

D2 · Edit / refine / reverse-engineer specs (brownfield + drift)

🔥adoption

Where: Spec Kit's #1 issue is literally "Can't Easily Update or Refine Existing Specs" + Reverse Engineering Command + Post-Implementation Debugging; OpenSpec /opsx:repair, hierarchical specs, multi-repo / monorepo; articles on spec durability & drift (Breunig's "triangle in sync").

Move: Wave 4 (living specs). Inline spec editing in the viewer, an /analyze-style drift/repair gate, and codebase reverse-engineering for adoption.

D3 · "Is SDD already obsolete?" — facts / BDD / IDSD debate

🔥adoption + content

Where: Kapil Ahuja's "SDD will collapse → IDSD/ICE" series vs Wasowski's "write facts, not specs" & "BDD is the missing link" vs Falk Gottlob's rebuttal "the prototype is the spec, the eval is the acceptance test — kill both." Papalini's 1.7×-more-issues stat. YouTube: "orchestration is just YOLO with better visuals."

Move: every voice lands on executable acceptance criteria / evals as the durable artifact — exactly Part 1's verified-completion + eval opportunity. Absorb the critique: gate mark-complete on real tests. And it's a great video.

D4 · Context rot / context durability

🔥adoption + competitive

Where: Rick Hightower's "context rot is killing your agents" series; Kapil's ICE; Wasowski's "managing agent context across the SDLC"; Goecke's "context engineering → SDD."

Move: the showdown article slots Spec Kit as the "spec layer" and GSD as the "context layer." Companion's .spec-context.json trace is a context-durability layer — position it as adding context survival on top of Spec Kit.

D5 · Governance & "approve intent, not code"

▲ risingadoption + competitive

Where: Papalini — governance is the new bottleneck once AI writes specs; devs shift to approving intent (and the study where they were ~19% slower while feeling faster).

Move: Companion's gates + trace + mark-complete are the oversight surface. Invest in approve/reject-with-reason; keep the trace honest about time.

D6 · Agent teams / multi-agent / worktrees

▲ risingadoption

Where: Superpowers' top open issue Agent Teams (21 comments); BMAD agent teams; Spec Kit multi-agent in one project + spawn worktree; cmux "Claude orchestrating Claude."

Move: Wave 5 fan-out + "compare implementations" worktree view. Worktree-per-branch is a small concrete win.

D7 · Lean vs ceremony; "plans over-specify"

▬ steadyadoption

Where: Superpowers' #2 open issue "plans over-specify, leaving no room for executor judgment"; iron-triangle "intent made the document irrelevant"; "GSD without ceremony"; "it looks like waterfall."

Move: validates fast-path-as-default; argues for gray-box plans (specify the interface, leave implementation judgment).

D8 · Vendor lock-in & artifact portability

▲ emergingcompetitive + content

Where: Kapil's "Five Dependency Layers" (lock-in relocates up to your spec format/toolchain) & "SDD will collapse on upstream change."

Move: plain-file trace + stock-Spec-Kit-shaped artifacts = anti-lock-in ("your specs are just files; swap the model underneath"). Risk to avoid: becoming a proprietary superset that adds a lock-in layer.

D9 · Autonomy / loops / "builds while you sleep"

▲ risingadoption + content

Where: Papalini's "Loop Engineering" & "builds while you sleep"; cmux hands-off, "24/7 agents," "Agentic OS."

Move: auto-mode (Wave 2) + token/cost budget with auto-pause.

D10 · Adoption scaffolding: templates, maturity, ROI

▬ steadycontent + adoption

Where: Wasowski's "templates + golden paths," "three maturity levels," "ROI (defend it to the board)."

Move: companion-template-profiles; a content gap for exec/ROI and maturity-ladder framings almost nobody covers.

Content / video ideas
#TitleWhy it lands
1"Is Spec-Driven Development Already Obsolete?"Rides the hottest debate (D3); steelman the critiques, land on evals as synthesis.
2"Brownfield SDD: Adopt Specs in an Existing Codebase"The single most-requested capability (D2); barely covered well anywhere.
3"Stop Trusting 'Done': Verified Completion w/ a Second-Model Review"Finger-guns + evals; shows cross-provider review live.
4"Which Model for Which Step?"Rides the Sonnet-5-vs-Opus comment war (D1); routing across a pipeline.
5"Spec Drift Is Killing Your Codebase"D2/D4; keep spec, code, tests in sync (Breunig's triangle).
6"One GUI, Every Provider"Companion demo framed against the "support my tool" flood (D1).
Competitive moves

Lead with cross-provider (D1) — every rival is ecosystem-locked and buried in "support X" issues; Companion's orchestration is the answer they can't easily ship.

Spec Kit users can't edit/refine specs (their #1 issue) → Companion's inline-edit viewer + adopt flow wins them. OpenSpec users want /repair + multi-repo specs → drift + scale. Superpowers users hit "plans over-specify" + slowness → lean routing + gray-box + opt-in.

Position as the composition / glue layer (D4) — the framework-showdown consensus is "combine layers, don't pick one." Companion orchestrates on top of Spec Kit; it's the one tool that composes.

This week's focus
  1. [A1] Make cross-provider the headline everywhere — README, listing, videos. #1 demand in every source. (competitive + adoption)
  2. [A2] Prioritize brownfield + spec-editing / drift (adopt flow, inline refine, /analyze repair) — Wave 4, most-requested capability. (adoption)
  3. [A3] Build verified-completion + optional eval layer — absorbs the "SDD is obsolete" critique. (adoption + content)
  4. [A4] Ship the two hottest videos — "Is SDD obsolete?" and "Brownfield SDD." (content)
  5. [A5] Reposition the trace as the context-durability + anti-lock-in layer on Spec Kit. (competitive + content)

This runs weekly now

Part 2 is the baseline for the new demand-radar skill — the same four lanes, scheduled Mondays. Every run refreshes the tracker and tags what's rising vs fading, so you watch the trend line, not one snapshot.

Full write-up: Part 2 (markdown) · tracker: demand-radar/_tracker.md · Part 1 — Video Trends