Skip to content

tvproductions/gzkit / .agents/skills/gz-competitor-radar

gz-competitor-radar

Verified

Use for the monthly competitor-discovery radar over spec-driven and agent-workflow tools.

gz-competitor-radar

Monthly competitor radar for spec-driven development, agent skills, executable spec ecosystems, and adjacent workflow tools.

Posture

  • Discovery: B+. Scan the seeded registry and actively discover new candidates. Admit new candidates only when the scan cites evidence and names a gzkit-relevant strength, rejection, or route.
  • Execution: B-. The report may recommend ADR/GHI/pool moves, but it does not mutate governance artifacts. It opens a grill/design discussion first.
  • Report structure: C. Render both product snapshots and strength-pattern maps. Products show trajectory; patterns prevent copying brands.
  • Source of truth. artifacts/reports/competitor-radar/registry.json and monthly scan JSON files govern the Markdown. Markdown reports under artifacts/reports/competitor-radar/ are generated projections and must not be hand-edited.
  • Portability. Deterministic scripts live with this skill under scripts/, per Agent Skills packaging guidance.
  • Cadence. The chore competitor-radar-currency puts the radar on a staleness clock (operator, 2026-10-26: "get it on a chore rotation (staleness) - yes, chore it"). Its maximum age is the chore's declared staleness.periodDays in .gzkit/chores/registry.json, and uv run gz chores status reports it overdue when the newest scan is older. The newest file under artifacts/reports/competitor-radar/scans/ shows when it last ran.

When To Use

  • Monthly competitor/status scan.
  • Ad-hoc scan when a new comparator appears.
  • Before deciding whether competitor strengths warrant a new pool ADR, existing pool update, active ADR amendment, GHI, or explicit rejection.
  • When the operator asks to grill competitor-derived recommendations.

Workflow

  1. Refresh sources. Browse current official sources first: project homepage, repository, docs, release notes, marketplace entry, and credible radar-style commentary. Do not rely on memory.

  2. Update scan JSON. For a new month, create the skeleton with radar.py new-scan --month YYYY-MM: it seeds one snapshot per registry competitor and refuses an existing month unless --overwrite is passed. Write findings to artifacts/reports/competitor-radar/scans/YYYY-MM.json. The agent authors this JSON from evidence and from grill answers. The operator does not edit JSON or Markdown.

  3. Render reports.

    uv run python .gzkit/skills/gz-competitor-radar/scripts/radar.py render
    
  4. Validate JSON and generated Markdown.

    uv run python .gzkit/skills/gz-competitor-radar/scripts/radar.py validate
    
  5. Open the grill discussion. For each suggested move, ask one question at a time, recommend an answer, and walk the decision tree. Record answers back into the scan JSON, rerender, and revalidate.

  6. Route only after decisions. Approved outcomes may update pool ADRs, file GHIs through ghi-author, or start design/promotion work. Rejections are recorded explicitly. A move's route must be one of existing-pool-adr, new-pool-adr, open-feature-adr, ghi, explicit-rejection or discussion-only; foundation ADRs are closed to new authoring (AGENTS.md § Gate Covenant), so there is no foundation route.

Grill Discipline

Ask the per-move questions in references/grill-questions.md for every suggested move, and its admission questions for every new candidate. Ask one question at a time. Prefer a recommended answer with tradeoffs. If the answer is discoverable from repo state or source evidence, gather it instead of asking the operator to type it.

Scripts

All scripts are stdlib Python and portable with the skill:

uv run python .gzkit/skills/gz-competitor-radar/scripts/new_scan.py --month YYYY-MM
uv run python .gzkit/skills/gz-competitor-radar/scripts/render_report.py
uv run python .gzkit/skills/gz-competitor-radar/scripts/validate_registry.py
uv run python .gzkit/skills/gz-competitor-radar/scripts/validate_report.py

The wrapper scripts delegate to scripts/radar.py. validate_registry.py and validate_report.py both run the same full validate.

Data Contract

Registry source:

artifacts/reports/competitor-radar/registry.json

Monthly scan source:

artifacts/reports/competitor-radar/scans/YYYY-MM.json

Generated reports:

artifacts/reports/competitor-radar/index.md
artifacts/reports/competitor-radar/YYYY-MM.md

Validation compares rendered Markdown byte-for-byte against the JSON source. If the Markdown differs, edit the JSON or the renderer, rerender, and revalidate.

References

  • references/registry.schema.json — registry shape.
  • references/scan.schema.json — monthly scan shape.

The schemas name only the top-level keys. radar.py validate is the authority: it also checks each competitor, pattern, snapshot and move, their ids and cross-references, and the route and watch-level vocabularies.

  • references/report-template.md — report section contract.
  • references/grill-questions.md — grill question bank.

Related Skills

  • gz-design
  • gz-adr-evaluate
  • gz-justify
  • ghi-author
  • gz-agent-sync
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall