Is mk:lazy-agent-loader safe to install?

SaferSkills independently audited mk:lazy-agent-loader (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.

CategoryWeightCategory scoreContribution

Securityprompt, exec, net, exfil, eval

35%

100

35.0 pts

Supply chainhash, typosquat, maintainer, lockfile

20%

100

20.0 pts

Maintenancestaleness, pinning, CI

15%

100

15.0 pts

TransparencySKILL.md, perms, README

15%

100

15.0 pts

Communityinstalls, verify, response

15%

100

15.0 pts

Findings & checks · 0 flagged

Securityscore 100 · 0 findings

✓—All security checks passedNo findings in this category for the latest scan.pass

Supply chainscore 100 · 0 findings

✓—All supply chain checks passedNo findings in this category for the latest scan.pass

Maintenancescore 100 · 0 findings

✓—All maintenance checks passedNo findings in this category for the latest scan.pass

Transparencyscore 100 · 0 findings

✓—All transparency checks passedNo findings in this category for the latest scan.pass

Communityscore 100 · 0 findings

✓—All community checks passedNo findings in this category for the latest scan.pass

The primary manifest — the file an agent reads to learn what this artifact does.

.claude/skills/lazy-agent-loader/SKILL.md · 1 file

.claude/skills/lazy-agent-loader/SKILL.md3.8 KB · Markdown

name: mk:lazy-agent-loader

description: Use when managing agent context budget. Loads agent definitions on-demand instead of all-at-once, reducing token consumption in multi-agent workflows.

triggers:

- when agent is selected

- agent:load

allowed-tools: Read, Glob

source: aura-frog

keywords:

- lazy-agent-loader

- context-budget

- on-demand-load

- agent-load-signal

- multi-agent

when_to_use: Auto-invoked when managing agent context budget — loads agent definitions on-demand. Not user-callable directly.

user-invocable: false

owner: research

criticality: medium

status: active

runtime: claude-code

Lazy Agent Loader

Priority: HIGH - Load agents on-demand

Purpose

Reduce token usage by:

Loading only agent summaries initially (~50 tokens each)
Loading full definition only when agent is activated
Caching loaded agents in session state

Agent Index (Summaries Only)

agent_index[15]{id,category,specialty,keywords}:
  orchestrator,infra,Task routing/complexity classification,route/classify/assign/complexity
  planner,planning,Two-lens planning/Gate 1,plan/feature/design/scope/requirements
  architect,planning,ADRs/system design,architecture/schema/api/infra/design/migration
  developer,dev,Implementation (TDD),implement/build/code/feature/fix/refactor
  tester,quality,TDD enforcement/test writing,test/testing/coverage/qa/jest/vitest/pytest
  reviewer,quality,5-dimension code review/Gate 2,review/audit/quality/standards
  security,quality,Security audit/BLOCK verdicts,security/vulnerability/audit/owasp/injection
  shipper,ops,Deploy pipeline/PR creation,ship/deploy/release/commit/pr/ci
  documenter,ops,Living documentation and changelogs,documentation:changelog:readme
  analyst,infra,Cost tracking/pattern extraction/memory,cost/budget/patterns/metrics/usage/memory
  researcher,planning,Technology research/library evaluation,research/compare/evaluate/library
  brainstormer,planning,Solution brainstorming/trade-offs,brainstorm/alternatives/tradeoff/explore
  journal-writer,ops,Failure documentation/lessons,failure/incident/postmortem/lesson
  git-manager,ops,Git operations/commit/push/PR,git/commit/push/pull/branch/pr/merge
  ui-ux-designer,dev,UI/UX design/frontend patterns,design/ui/ux/css/layout/accessibility/responsive

Loading Strategy

Initial Load (~1200 tokens)

1. Load this index file (agent_index)
2. DO NOT load individual agent files
3. Use index for agent detection scoring

On Agent Selection (~500-2000 tokens per agent)

1. Agent scores >=80 (PRIMARY) → Load full definition
2. Agent scores 50-79 (SECONDARY) → Load summary only
3. Agent scores <50 (OPTIONAL) → Don't load

Full Definition Location

.claude/agents/[agent-name].md

Integration with Agent Detector

Step 1: Score agents using agent_index keywords
Step 2: Identify PRIMARY agent(s) with score >=80
Step 3: Load ONLY PRIMARY agent full definitions
Step 4: For SECONDARY agents, use summary from index

References

Reference	When to load	Content
[loading-details.md](./references/loading-details.md)	When loading specific agents	Commands, token savings, cache strategy, examples

Note: This skill is automatically used by agent-detector for optimized loading.

Gotchas

Agent loaded too late missing context: Lazy loading skips context that was available at session start → Pre-load agents that need session-start context (orchestrator, analyst)
Cache serving stale agent definition: Agent file updated but cached version used → Invalidate cache on file mtime change

mk:lazy-agent-loader— agent skill

Is mk:lazy-agent-loader safe to install?

Lazy Agent Loader

Purpose

Agent Index (Summaries Only)

Loading Strategy

Initial Load (~1200 tokens)

On Agent Selection (~500-2000 tokens per agent)

Full Definition Location

Integration with Agent Detector

References

Gotchas

Audit the pieces. Scan the whole. Decide.