skills-workflows_skill

This skill designs multi-skill workflow pipelines with artifact-based state handoff to streamline planning, implementation, and review.
  • TypeScript

25

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill outfitter-dev/agents --skill skills-workflows

  • SKILL.md7.1 KB

Overview

This skill designs multi-skill workflow systems that chain individual skills via artifact-based state handoff. It prescribes shared file locations, artifact formats, fork vs inherit context patterns, and preprocessing rules to make pipelines deterministic and auditable. Use it to standardize pipelines from triage through review and to reduce context pollution and lost state.

How this skill works

The skill defines conventions: where to store artifacts, how each step reads prior artifacts and writes its own, and what shared files (context.md, constraints.md) must contain. It documents fork vs inherit usage to control conversation noise, sets rules for preprocessing commands that run before model input, and prescribes guardrails for side-effect skills. The outcome is a sequenced pipeline where every transition is a file-based artifact consumed by the next step.

When to use it

  • Building multi-skill pipelines (triage → plan → implement → review)
  • Designing deterministic state handoff between workflow steps using artifacts/
  • Setting up shared conventions like .claude/skills/_shared/context.md and constraints.md
  • Implementing preprocessing with !command syntax to inject deterministic snapshots
  • Choosing fork vs inherit to isolate analysis from implementation

Best practices

  • Store all transient and persistent step outputs in artifacts/ so state is file-backed and auditable
  • Keep .claude/skills/_shared/context.md updated with decisions to avoid losing intent in conversation
  • Fork analysis and review skills (context: fork) to prevent context pollution; inherit for implementation when history is needed
  • Use disable-model-invocation: true for skills that cause side effects so they require explicit user invocation
  • Minimize allowed-tools per skill to the smallest set required to limit permission creep

Example use cases

  • Triage skill reads issue text, writes artifacts/triage.md with scope, risks, and acceptance criteria
  • Plan skill forks, reads artifacts/triage.md, produces artifacts/plan.md with tasks and estimated effort
  • Implement skill inherits context, reads artifacts/plan.md, updates code and .claude/skills/_shared/context.md with decisions
  • Test skill runs deterministic scripts (scripts/run-tests.sh) and writes artifacts/test-report.md for review
  • Review skill forks, aggregates artifacts/* and writes artifacts/review-notes.md for stakeholders

FAQ

Make each artifact self-contained and parseable: header, problem statement, scope, acceptance criteria, risks, and a generated-by line with timestamp.

When should I preprocess with !command?

Preprocess when you need deterministic snapshots (git status, PR diffs, schema queries) but avoid preprocessing large outputs, dynamic queries, or commands with side effects.

When to choose fork vs inherit?

Use fork for clean-room analysis, parallel work, and reviews. Use inherit when the skill needs prior conversation or decisions to implement changes.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational
skills-workflows skill by outfitter-dev/agents | VeilStrat