mirror of https://github.com/alirezarezvani/claude-skills.git synced 2026-10-06 02:50:08 +00:00

feat(agents,commands): trigger descriptions + gated workflow upgrades

- 17 root agents gained 'Use when...' trigger descriptions (<=1024 chars);
  cs-product-analyst placeholder body rewritten around real tools
- 21 commands had invalid YAML frontmatter fixed (unquoted inner colons);
  14 commands gained argument-hint; 6 duplicate commands marked with
  canonical-copy sync comments
- /prd: forcing questions + refuse-to-draft gate + required-sections checklist
- /sprint-plan: capacity math + carry-over check + definition-of-ready gate
- /tdd: rebuilt around the real tdd-guide assets with smoke-tested snippets
- cs-senior-engineer repointed from retired release-manager to
  changelog-generator

https://claude.ai/code/session_019AJddAL1NADWMXsy1qNPQF

2026-06-11 03:58:42 +00:00

3.2 KiB

Raw Permalink Blame History

name	description
focused-fix	Deep-dive feature repair — systematically fix an entire feature/module across all its files and dependencies. Usage: /focused-fix <feature-path>

/focused-fix

Systematically repair an entire feature or module using the 5-phase protocol. Target: $ARGUMENTS (a feature path or module name).

If $ARGUMENTS is empty, ask the user which feature/module to fix.

Protocol — Execute ALL 5 Phases IN ORDER

Phase 1: SCOPE — Map the Feature Boundary

Identify the primary folder/files for the target feature
Read EVERY file in that folder — understand its purpose
Create a feature manifest:

FEATURE SCOPE:
  Primary path: <path>
  Entry points: [files imported by other parts of the app]
  Internal files: [files only used within this feature]
  Total files: N

Phase 2: TRACE — Map All Dependencies

INBOUND (what this feature imports):

For every import statement, trace to source, verify it exists and is exported
Check env vars, config files, DB models, API endpoints, third-party packages

OUTBOUND (what imports this feature):

Search entire codebase for imports from this feature
Verify consumers use correct API/interface

Output a dependency map with inbound, outbound, env vars, and config files.

Phase 3: DIAGNOSE — Find Every Issue

Run ALL diagnostic checks:

Code: imports resolve, no circular deps, types consistent, error handling, TODO/FIXME
Runtime: env vars set, migrations current, API shapes correct
Tests: run ALL related tests, record failures, check coverage
Logs: check git log for recent changes, search error logs
Config: validate config files, check dev/prod mismatches

For each issue found:

Confirm root cause with evidence before adding to fix list
Assign risk: HIGH (public API, auth, >3 callers) / MED (internal with tests) / LOW (leaf module)

Output a diagnosis report with issues grouped by severity.

Phase 4: FIX — Repair Systematically

Fix in this EXACT order:

Dependencies — broken imports, missing packages
Types — type mismatches at boundaries
Logic — business logic bugs
Tests — fix or create tests for each fix
Integration — verify end-to-end with consumers

Rules:

Fix ONE issue at a time, run related test after each
If a fix breaks something else → go back to DIAGNOSE
Fix HIGH before MED before LOW
3-Strike Rule: If 3+ fixes create NEW issues, STOP. Tell the user the architecture may need rethinking, not patching.

Phase 5: VERIFY — Confirm Everything Works

Run ALL tests in the feature folder
Run ALL tests in files that import from this feature
Run full test suite if available
Summarize all changes made

Output a completion report with files changed, fixes applied, test results, and consumers verified.

Iron Law

NO FIXES WITHOUT COMPLETING SCOPE → TRACE → DIAGNOSE FIRST

If you haven't finished Phase 3, you cannot propose fixes.

engineering/focused-fix — Full SKILL.md with detailed checklists, output templates, and anti-patterns
superpowers:systematic-debugging — For individual complex bugs found during Phase 3

3.2 KiB Raw Permalink Blame History