AgentStack
SKILL verified Apache-2.0 Self-run

Radar Suite

skill-terryc21-radar-suite-radar-suite · by Terryc21

Unified entry point for the 6-skill radar family (5 companions + capstone). Routes to individual skills, runs targeted (Tier 2) or full (Tier 3) audit sequences, owns the unified ledger, and provides cross-skill operations (status, verify, ledger, link, deferred, fresh/no-fresh). Triggers: "radar suite", "full audit", "run all radars", "/radar-suite".

No reviews yet
0 installs
13 views
0.0% view→install

Install

$ agentstack add skill-terryc21-radar-suite-radar-suite

✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

Are you the author of Radar Suite? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Radar Suite — Unified Entry Point

> Single command to run any radar skill or the full audit sequence.

Quick Commands

Tier Selection (see radar-suite-core.md Tier System for full spec)

| Command | Tier | Description | |---------|------|-------------| | /radar-suite | 1 | Interactive menu (default: single skill) | | /radar-suite [skill] | 1 | Run specific skill (data-model, time-bomb, roundtrip, ui-path, ui-enhancer, capstone) | | /radar-suite --skills dmr,tbr | 2 | Targeted: run 2-3 skills with cross-skill handoffs | | /radar-suite --changed | 2 | Targeted: auto-select skills from git diff | | /radar-suite --changed --since YYYY-MM-DD | 2 | Targeted: auto-select from changes since date | | /radar-suite --scope [path] | 2 | Targeted: restrict audit to a directory subtree | | /radar-suite --full | 3 | Full pipeline: all 6 skills + capstone + UX enhancements | | /radar-suite full | 3 | Alias for --full (backward compatible) |

State Flags (combine with any command above)

| Flag | Description | |---------|-------------| | --fresh | Archive prior ledger.yaml and known-intentional.yaml before this run; treat the codebase as never-before-audited. See "Fresh Scan Check" section. | | --no-fresh | Skip the fresh-scan prompt; always continue with existing state. Useful for scripted/CI runs. |

Management Commands

| Command | Description | |---------|-------------| | /radar-suite status | Show audit progress across all skills | | /radar-suite resume | Resume from last checkpoint | | /radar-suite ledger | View unified finding ledger with optional filters | | /radar-suite ledger --open | Show open findings only | | /radar-suite ledger --deferred | Show deferred findings only | | /radar-suite ledger --impact crash | Filter by impact category (crash, data-loss, ux-broken, ux-degraded, polish, hygiene) | | /radar-suite ledger --skill [name] | Filter by source skill | | /radar-suite deferred | Generate DEFERRED.md from ledger | | /radar-suite verify | Re-verify all fixed findings | | /radar-suite verify RS-NNN | Re-verify a single finding | | /radar-suite verify --changed | Re-verify only findings in changed files | | /radar-suite link RS-NNN --root-cause-of RS-NNN [RS-NNN...] | Link findings as root cause/symptom | | /radar-suite link RS-NNN --duplicate-of RS-NNN | Mark finding as duplicate | | /radar-suite link RS-NNN --supersedes RS-NNN | Mark finding as superseding another | | /radar-suite bug-echo RS-NNN | Manually trigger the Bug-Echo Handoff prompt for a finding (bypasses Open→Fixed gate). See Bug-Echo Handoff in radar-suite-core.md. |

Available Skills

| Skill | Purpose | Est. Time | |-------|---------|-----------| | data-model-radar | Audit @Model layer for completeness, serialization, relationships | ~30-60 min | | time-bomb-radar | Find deferred operations that crash on aged data | ~15-25 min | | roundtrip-radar | Trace user workflows end-to-end for data safety | ~20-40 min | | ui-path-radar | Find dead ends, broken promises, navigation issues | ~15-30 min | | ui-enhancer-radar | Visual UI audit with design intent interview | ~20-45 min | | capstone-radar | Aggregate grades, ship/no-ship decision | ~15-30 min |

Count vocabulary used throughout this spec

To prevent ambiguity, this document uses three distinct counts:

| Term | Meaning | Skills included | |------|---------|-----------------| | 5 companions | The audit skills that capstone consumes | data-model + time-bomb + roundtrip + ui-path + ui-enhancer | | 6 skills / the family | All audit + aggregation skills | 5 companions + capstone | | 2-5 skills (used in --skills validation) | Subset accepted by Tier 2 routing | Any 2-5 of the 5 companions (capstone excluded — runs after) |

Throughout the rest of this spec, references to "5 companions" or "5 skills" mean companion-only (audit skills before capstone aggregates them). References to "6 skills" mean the complete family. References to "5-skill" (without "companion") elsewhere in this document are historical drift and should read "6-skill" — see capstone-radar's matching framing.


Session Setup (MANDATORY -- runs before anything else)

On EVERY /radar-suite invocation, check .radar-suite/session-prefs.yaml:

If file exists AND has experience_level set: Show stored preferences and ask [Enter to continue] or type "change" to adjust settings. Then proceed to Stale-Deferred Check, then Interactive Menu.

If file does not exist OR experience_level is missing: Run full setup below before presenting the menu, fix timing, or any audit work. Do NOT skip this. Do NOT jump to exploration.

Setup Questions (single AskUserQuestion, 4 questions)

Q1 — Experience level?

  • Experienced (Recommended) -- Concise, no definitions
  • Senior/Expert -- Terse, file:line only
  • Intermediate -- Standard terms, explain non-obvious
  • Beginner -- Plain language, define terms

Q2 — Table format?

  • Full tables (Recommended) -- 9-column Issue Rating Tables at every decision point
  • Compact tables -- 3-column with details below
  • No tables -- Skip rating tables entirely (findings listed as text)

Q3 — Fix handling?

  • Auto-fix safe items (Recommended) -- Apply isolated, low-blast-radius fixes automatically
  • Review first -- Present all findings, approve each wave
  • Batch mode -- Approve all fixes in each wave at once

Q4 — Explain what this skill does?

  • No, let's go (Recommended) -- Skip explanation
  • Yes, briefly -- 3-5 sentence explanation

Store answers in .radar-suite/session-prefs.yaml as experience_level, table_format, fix_mode (3 of the 4 questions — Q4's explain toggle is a one-time per-session choice and is NOT persisted). The fix_timing key is added later by the Fix Timing section below (after the Interactive Menu); see § Fix Timing for its values. See radar-suite-core.md for experience-level output rules.

session-prefs.yaml schema (consolidated reference)

The full set of keys radar-suite writes to .radar-suite/session-prefs.yaml:

| Key | Source | Values | |---|---|---| | experience_level | Q1 above | beginner / intermediate / experienced / senior | | table_format | Q2 above | full / compact / none | | fix_mode | Q3 above | auto / review / batch | | fix_timing | § Fix Timing below | recommended / all_per_skill / all_after_capstone | | tier | Menu routing / --full / --skills | 1 / 2 / 3 | | tier_skills | --skills invocation | Array of abbreviations (only when tier: 2) | | dippy_check | radar-suite-core.md § Environment Pre-flight | Object with path_has_spaces, dippy_installed, checked_on | | last_skill / last_session | Updated on session end | Skill name / ISO 8601 timestamp | | accepted_risks | User opt-in per finding | Array of RS-NNN IDs |

Execution Order

The invocation flow is strictly:

  1. Session Setup (this section) -- configure or confirm preferences
  2. Write Execution Rules Memory -- write/update the auto-generated memory (see below)
  3. Fresh Scan Check -- offer to archive prior state and start clean
  4. Stale-Deferred Check -- check for overdue findings
  5. Fix Timing -- ask when fixes should be applied (fresh audits only)
  6. Interactive Menu -- choose what to audit
  7. Skill execution -- run the selected skill(s)

Never skip steps 1-5. Never start exploration or scanning before completing them.

Write Execution Rules Memory (MANDATORY -- after session setup)

After session setup completes (step 1), write or update a memory file that captures the cross-cutting rules most commonly violated during long sessions. This memory survives context compression and is re-loaded from disk on every session.

File: Write to the project's auto-memory directory (the same location used by the auto memory system). The file name is radar_suite_execution_rules.md.

Content (write exactly this, substituting session prefs):

---
name: Radar Suite execution rules
description: Auto-generated rules for radar-suite display and behavior. Re-read before every AskUserQuestion.
type: feedback
---
1. 9-column table at every decision point: #, Finding, Urgency, Risk:Fix, Risk:No Fix, ROI, Blast Radius, Fix Effort, Status. Omit Status on first display only. Include on every re-display. Skip if TABLE_FORMAT = none.
2. Finding IDs always include short_title: RS-NNN (short description), never bare RS-NNN.
3. Progress banner after every wave, phase, or commit. Never leave a blank prompt. Always follow with the rating table (per rule 1), then AskUserQuestion.
4. Recommend fixing over deferring when effort = HIGH`
- `fix_effort` is `trivial` or `small`
- `blast_radius  Export has no back button (ui-path-radar) — fixed 2026-04-09

POLISH ([N] findings)
  ...

Options:
1. **Filter** — apply additional filters
2. **Details [RS-NNN]** — show full finding details with history
3. **Fix [RS-NNN]** — mark a finding as fixed
4. **Defer [RS-NNN]** — defer a finding with reason and review date
5. **Done** — exit ledger view

Finding Details

/radar-suite ledger RS-001 shows full details:

RS-001 [HIGH] CSV export drops Room and UPC columns on import
  Status: open
  Impact: data-loss
  Source: roundtrip-radar (also flagged by: data-model-radar)
  File: Sources/Managers/CSVManager.swift:142
  Confidence: verified
  Evidence: grep confirmed importCSV reads 25 fields, exportCSV writes 27
  Discovered: 2026-04-08
  Related: RS-007, RS-011

  History:
    2026-04-08 discovered by roundtrip-radar
    2026-04-08 also flagged by data-model-radar

No Ledger

If .radar-suite/ledger.yaml doesn't exist:

No finding ledger found. Run an audit skill to create one.
Recommended: /radar-suite full (run all skills) or /radar-suite data-model (start with foundation)

Handoff Flow

Each skill writes .radar-suite/[skill]-handoff.yaml on completion.

Capstone-radar reads all handoffs to:

  1. Aggregate findings
  2. Detect cross-skill patterns
  3. Produce unified grades
  4. Make ship/no-ship recommendation

Shared Patterns

See radar-suite-core.md for: Experience-Level Output Rules, Session Persistence, Checkpoint & Resume, Accepted Risks, Wave-Based Fix Presentation, Table Format, Issue Rating Tables, Unified Finding Ledger Protocol.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet — be the first.

Versions

  • v0.1.0 Imported from the upstream source.