Skip to content

im-shashanks/claude-plugins / shaktra/skills/shaktra-reference

shaktra-reference

Verified

Shared constants, quality standards, and state schemas for all Shaktra agents. Defines severity taxonomy, quality principles, guard tokens, review dimensions, and YAML schemas.

schemas/settings-schema.md

View on GitHub
# Settings Schema

Defines `.shaktra/settings.yml` — the single configuration file for all Shaktra thresholds and project metadata. Every threshold in the framework reads from this file; none are hardcoded.

## Schema

```yaml
project:
  name: string            # required — project display name
  type: string            # "greenfield" | "brownfield"
  language: string        # "python" | "typescript" | "go" | "java" | "rust" | etc.
  architecture: string    # "layered" | "hexagonal" | "clean" | "mvc" | "feature-based" | "event-driven" | ""
  test_framework: string  # "pytest" | "jest" | "vitest" | "go test" | etc.
  coverage_tool: string   # "coverage" | "istanbul" | "c8" | etc.
  package_manager: string # "pip" | "npm" | "pnpm" | "cargo" | etc.

tdd:
  coverage_threshold: integer       # default: 90 — Medium tier target
  hotfix_coverage_threshold: integer # default: 70 — Trivial tier target
  small_coverage_threshold: integer  # default: 80 — Small tier target
  large_coverage_threshold: integer  # default: 95 — Large tier target

quality:
  p1_threshold: integer  # default: 2 — max P1 findings before merge block

review:
  verification_test_persistence: string  # default: "ask" — "auto" | "always" | "never" | "ask"
  min_verification_tests: integer        # default: 5 — minimum independent verification tests

analysis:
  summary_token_budget: integer  # default: 600 — max tokens per artifact summary section
  incremental_refresh: boolean   # default: true — enable checksum-based incremental refresh

refactoring:
  safety_threshold: integer              # default: 80 — min coverage % before targeted refactoring
  structural_safety_threshold: integer   # default: 90 — min coverage % before structural refactoring
  max_characterization_tests: integer    # default: 20 — max tests added during FORTIFY phase

sprints:
  enabled: boolean              # default: true
  velocity_tracking: boolean    # default: true
  sprint_duration_weeks: integer # default: 2
  default_velocity: integer     # default: 15 — story points per sprint, used when no velocity history exists

pm:
  default_framework: string           # default: "rice" — "rice" | "weighted" | "moscow"
  quick_win_effort_threshold: integer # default: 3 — max story points for Quick Win classification
  big_bet_impact_threshold: integer   # default: 7 — min impact score for Big Bet classification
  min_persona_evidence: integer       # default: 2 — minimum evidence entries per persona
  min_journey_stages: integer         # default: 3 — minimum stages in a journey map

memory:
  confidence_start: float          # default: 0.6 — initial confidence for new entries
  confidence_reinforce: float      # default: 0.08 — confidence boost per reinforcing observation
  confidence_weaken: float         # default: 0.08 — confidence reduction per weakening observation
  confidence_contradict: float     # default: 0.20 — confidence reduction per contradicting observation
  confidence_archive: float        # default: 0.2 — entries below this are archived
  max_principles: integer          # default: 200 — max active principles
  max_anti_patterns: integer       # default: 100 — max active anti-patterns
  max_procedures: integer          # default: 50 — max active procedures
  max_observations_per_story: integer # default: 30 — max observations per story
  briefing_confidence_threshold: float # default: 0.4 — min confidence for briefing inclusion
  retrieval_tier1_max: integer         # default: 100 — max entries for inline briefing (Tier 1)
  retrieval_tier2_max: integer         # default: 500 — max entries for single-agent retrieval (Tier 2)
  max_briefing_entries: integer        # default: 15 — max entries in a generated briefing
  retrieval_chunk_size: integer        # default: 150 — entries per chunk in Tier 3 retrieval
```

## Consumer Reference

| Setting | Read By |
|---|---|
| `project.name` | init skill (writes), all agents (project identification) |
| `project.type` | analyze skill (brownfield guard), all agents (context) |
| `project.language` | analyze skill (glob patterns), general skill (domain context), all agents (context) |
| `project.architecture` | architect (validates alignment), sw-engineer (pattern selection), developer (code structure), sw-quality (ARC checks — conditional enforcement) |
| `project.test_framework` | all agents (implicit context — determines test syntax and runner conventions) |
| `project.coverage_tool` | all agents (implicit context — determines coverage report format) |
| `project.package_manager` | all agents (implicit context — determines install/build commands) |
| `tdd.coverage_threshold` | developer, story-tiers gate matrix |
| `tdd.hotfix_coverage_threshold` | developer, story-tiers gate matrix |
| `tdd.small_coverage_threshold` | developer, story-tiers gate matrix |
| `tdd.large_coverage_threshold` | developer, story-tiers gate matrix |
| `quality.p1_threshold` | sw-quality, code-reviewer, severity-taxonomy merge gate |
| `review.verification_test_persistence` | code-reviewer |
| `review.min_verification_tests` | code-reviewer |
| `analysis.summary_token_budget` | cba-analyzer (via analysis-output-schemas.md budgets), analyze skill (validation) |
| `analysis.incremental_refresh` | analyze skill (refresh workflow) |
| `refactoring.safety_threshold` | refactoring-pipeline (orchestrator), test-agent (characterization mode) |
| `refactoring.structural_safety_threshold` | refactoring-pipeline (orchestrator), test-agent (characterization mode) |
| `refactoring.max_characterization_tests` | test-agent (characterization mode) |
| `sprints.enabled` | scrummaster, tdd-pipeline (orchestrator), workflow-template (orchestrator) |
| `sprints.velocity_tracking` | scrummaster |
| `sprints.sprint_duration_weeks` | scrummaster |
| `sprints.default_velocity` | scrummaster (fallback when no velocity history) |
| `pm.default_framework` | product-manager (prioritize mode), pm skill (prioritization workflow) |
| `pm.quick_win_effort_threshold` | product-manager (RICE classification) |
| `pm.big_bet_impact_threshold` | product-manager (RICE classification) |
| `pm.min_persona_evidence` | product-manager (persona-create validation) |
| `pm.min_journey_stages` | product-manager (journey-create validation) |
| `memory.confidence_start` | memory-curator (new entry creation) |
| `memory.confidence_reinforce` | memory-curator (reinforcement) |
| `memory.confidence_weaken` | memory-curator (weakening) |
| `memory.confidence_contradict` | memory-curator (contradiction) |
| `memory.confidence_archive` | memory-curator (archival threshold) |
| `memory.max_principles` | memory-curator (rotation limit) |
| `memory.max_anti_patterns` | memory-curator (rotation limit) |
| `memory.max_procedures` | memory-curator (rotation limit) |
| `memory.max_observations_per_story` | all observation-writing agents (cap check) |
| `memory.briefing_confidence_threshold` | orchestrators (inline briefing), memory-retriever |
| `memory.retrieval_tier1_max` | orchestrators via `memory_retrieval.py` |
| `memory.retrieval_tier2_max` | orchestrators via `memory_retrieval.py` |
| `memory.max_briefing_entries` | orchestrators (inline briefing), memory-retriever |
| `memory.retrieval_chunk_size` | `memory_retrieval.py` (Tier 3 splitting) |

## Environment Variable Overrides

Hook scripts support these environment variables for override scenarios (CI, testing, emergency fixes):

| Variable | Default | Purpose |
|---|---|---|
| `SHAKTRA_ALLOW_MAIN_BRANCH` | unset | Set to any value to bypass protected branch hook |
| `SHAKTRA_SKIP_P0_CHECK` | unset | Set to any value to bypass P0 finding check at completion |
| `CLAUDE_PROJECT_DIR` | `cwd` | Override project root for hook file resolution |

These are escape hatches — not regular workflow configuration. Do not set them permanently.
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