agentleFS
fierc3fierc3/agent-harness-template/.agents/skills/project-kickoff/SKILL.md

project-kickoff

Skill0 starsChanged 3 months agoMITView on GitHub

Use this when the opening DO KICKOFF section in AGENTS.md shows that a template has not been initialized.

14 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

16 rules. 0 are common, 5 are distinctive, and 11 are very specific to this project.

  • 0

    Common

    Most of the 13 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 11

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run this skill when AGENTS.md starts with a ## DO KICKOFF section.

    line 14
  2. Distinctive:

    Ask one concise batch of only the questions still needed to establish durable context:

    line 28
  3. Distinctive:

    Do not assume an answer when a choice materially affects the project.

    line 47
  4. Very specific:

    Do not complete kickoff with the template's MIT license merely by omission: obtain an explicit decision about the LICENSE file first.

    line 49
  5. Very specific:

    Only after the owner explicitly confirms the license, update LICENSE with the chosen license text, retain it when they explicitly choose …

    line 61
  6. Very specific:

    Use the skill-creator skill to add a focused skill only when it has owned, reusable guidance:

    line 70
  7. Very specific:

    Create tech-stack when framework, runtime, dependencies, or architecture need project-specific conventions beyond the shared context.

    line 73
  8. Very specific:

    Create ui-ux when UI work is in scope and durable design or accessibility guidance exists.

    line 75
  9. Very specific:

    Create testing when the project has established verification commands, test strategy, fixtures, coverage rules, or end-to-end workflow.

    line 77
  10. Very specific:

    Create deployment when deployment is in scope and there is a repeatable environment, release, secrets, CI/CD, or rollback workflow.

    line 79
  11. Distinctive:

    Do not create placeholder testing or deployment skills merely because those topics were discussed.

    line 82
  12. Distinctive:

    After the confirmed context and justified skills are written:

    line 88
  13. Distinctive:

    Do not leave template or kickoff language in the replacement README.

    line 93
  14. Very specific:

    Remove the complete ## DO KICKOFF section from AGENTS.md, while keeping its completed project context and all other durable instructions.

    line 94
  15. Very specific:

    Delete .agents/skills/project-kickoff/ as the final content change.

    line 96
  16. Very specific:

    Run ./scripts/verify-harness.sh after deleting the skill, when the script exists.

    line 98

“Common” is measured: kept by at least 6 in 10 of the 13 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Project kickoff

Turn this template into a project-specific workspace through a short, evidence-based discovery process. The goal is to record confirmed durable context, not to design the project or fill the repository with empty skills.

When to run

Run this skill when AGENTS.md starts with a ## DO KICKOFF section. Finish the kickoff before substantial implementation. Once that section is gone, this skill is no longer needed and must not be recreated unless the user asks for a new kickoff process.

1. Inspect before asking

Inspect the repository for existing code, manifests, documentation, CI, configuration, and design assets. Treat that evidence as authoritative. Do not ask questions that the repository already answers, and do not overwrite an existing codebase's conventions.

2. Ask focused questions

Ask one concise batch of only the questions still needed to establish durable context:

  • Is this a new project or an existing-codebase migration?
  • What problem, users, and primary outcome does the project have?
  • What license should apply? For a new project, explicitly confirm whether to keep this template's MIT license, choose another license, or use no license. For an existing project, confirm the current license and ownership before changing it.
  • For a new project: which language, runtime, framework, package manager, and major services should be used? For an existing project: confirm discoveries and ask only where choices are unclear.
  • What commands should verify changes: tests, linting, formatting, type checks, builds, or end-to-end checks?
  • Is deployment in scope? If so, what environment, platform, CI/CD workflow, secrets policy, release process, and rollback expectations apply?
  • Is UI/UX in scope? If so, what design system, visual references, accessibility requirements, breakpoints, and interaction guidelines apply?

Do not assume an answer when a choice materially affects the project. If the user defers a decision, record it as undecided rather than inventing a default. Do not complete kickoff with the template's MIT license merely by omission: obtain an explicit decision about the LICENSE file first.

3. Record the shared project context

Replace the placeholder in AGENTS.md with a concise project description, confirmed architecture and conventions, and the canonical verification commands when known. Keep broad, durable rules in AGENTS.md; do not put detailed procedures there.

4. Apply the license decision

Only after the owner explicitly confirms the license, update LICENSE with the chosen license text, retain it when they explicitly choose MIT, or remove it only when they explicitly choose no license. Preserve an existing project's license unless the owner directs a change. Include the resulting license in the project README. Do not provide legal advice or choose a license on the owner's behalf.

5. Create only justified skills

Use the skill-creator skill to add a focused skill only when it has owned, reusable guidance:

  • Create tech-stack when framework, runtime, dependencies, or architecture need project-specific conventions beyond the shared context.
  • Create ui-ux when UI work is in scope and durable design or accessibility guidance exists.
  • Create testing when the project has established verification commands, test strategy, fixtures, coverage rules, or end-to-end workflow.
  • Create deployment when deployment is in scope and there is a repeatable environment, release, secrets, CI/CD, or rollback workflow.

Do not create placeholder testing or deployment skills merely because those topics were discussed. Each created skill must have a specific trigger, confirmed instructions, and verification steps.

6. Finish and clean up

After the confirmed context and justified skills are written:

  1. Replace README.md entirely with concise, project-specific documentation: purpose, setup, development and verification commands, deployment guidance when applicable, and any user-facing usage instructions that are known. Do not leave template or kickoff language in the replacement README.
  2. Remove the complete ## DO KICKOFF section from AGENTS.md, while keeping its completed project context and all other durable instructions.
  3. Delete .agents/skills/project-kickoff/ as the final content change. This skill is temporary scaffolding, not part of the initialized project.
  4. Run ./scripts/verify-harness.sh after deleting the skill, when the script exists.

Summarize the decisions recorded, files created or removed, and important unanswered questions. Verify that every remaining skill has valid frontmatter, a description starting with Use this when, and fewer than 200 lines.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.