agentleFS
clarknova99clarknova99/multi-provider-harness/.agents/skills/executing-plans/SKILL.md

executing-plans

Skill1 starsChanged 4 months agoMITView on GitHub

How to execute a written implementation plan in docs/plans/ task-by-task — subagent-driven where independent, one commit per task, the verify gate green before each commit, and the cross-provider checkpoint after each. Use whenever you have an approved plan to implement (Status DRAFT/IN_PROGRESS), when the user says "start the plan", "keep going", "implement task N", or when resuming interrupted plan work.

189 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use chinese-code-review, chinese-commit-conventions and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • Leodorareluctant259Skill

    chinese-code-review

    中文代码审查规范——在保持专业严谨的同时,用符合国内团队文化的方式给出有效反馈

    Also in 8 projects that use this

  • Leodorareluctant259Skill

    chinese-commit-conventions

    中文 Git 提交规范 — 适配国内团队的 commit message 规范和 changelog 自动化

    Also in 8 projects that use this

  • obraSkill

    diagnosing-superpowers

    Use 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.

    Also in 20 projects that use this

  • allanninalSkill

    algorand-vulnerability-scanner

    Scans Algorand smart contracts for 11 common vulnerabilities including rekeying attacks, unchecked transaction fees, missing field validations, and access control issues. Use when auditing Algorand projects (TEAL/PyTeal). (project, gitignored)

    Also in 4 projects that use this

  • allanninalSkill

    cairo-vulnerability-scanner

    Scans Cairo/StarkNet smart contracts for 6 critical vulnerabilities including felt252 arithmetic overflow, L1-L2 messaging issues, address conversion problems, and signature replay. Use when auditing StarkNet projects. (project, gitignored)

    Also in 4 projects that use this

  • farslabSkill

    aws-agentic-ai

    AWS Bedrock AgentCore comprehensive expert for deploying and managing AI agents at scale. Use when working with any AgentCore service including Gateway, Runtime, Memory, Identity, Code Interpreter, Browser, Observability, Agent Registry, or Evaluations. Covers agent deployment, MCP tool integration, credential management, agent discovery, governance workflows, and automated quality assessment. Essential when user mentions AgentCore, agent runtime, agent registry, agent evaluation, MCP gateway, deploy agent, register MCP server, discover agents, evaluate agent quality, agent credentials, or wants to build, deploy, catalog, or monitor AI agents on AWS.

    Also in 3 projects that use this

  • +3 more from Leodorareluctant259
  • +1 more from allanninal

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 178 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

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

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

    Run the resume/surfacing check in checkpointing-progress: read docs/state/CURRENT.md, reconcile with git status / git log, and confirm …

    line 10
  2. Distinctive:

    Scope it: Read the task's Files / Interfaces / Steps.

    line 14
  3. Very specific:

    Build it test-first: Follow test-driven-development: failing test (red) → least code (green) → refactor.

    line 16
  4. Distinctive:

    After the last task, review the entire branch diff, not just the per-task diffs — cross-task seams hide bugs no single task's review could …

    line 22

“Common” is measured: kept by at least 6 in 10 of the 178 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

Skill: Executing plans (subagent-driven, checkpointed)

You have a written plan in docs/plans/ with checkbox tasks. Execute it task by task — never all at once, never skipping the per-task gate.

Before you start

Run the resume/surfacing check in checkpointing-progress: read docs/state/CURRENT.md, reconcile with git status / git log, and confirm which task is next. If a *-progress.md ledger disagrees with the tree, trust the tree and reconcile the ledger first.

The per-task loop

For each task, in order:

  1. Scope it. Read the task's Files / Interfaces / Steps. It should be one coherent, committable unit. If it's really several, do them as sub-steps but still commit once per task.
  2. Dispatch. Independent tasks (no shared files, no ordering dependency) can each run in their own subagent — spawn them in parallel, handing each the task text plus the plan's Global Constraints. Tasks that share files or must be sequenced run in order in the main line. Never parallelize work that touches the same file.
  3. Build it test-first. Follow test-driven-development: failing test (red) → least code (green) → refactor. UI-facing tasks also add whatever integration/UI coverage the plan names.
  4. Verify. The project's verify gate must be green. If the plan explicitly allows an intentional red mid-refactor, say so in the ledger.
  5. Review. A code-reviewer-persona pass on the task's diff before it's called done; fix findings or record why they're deferred.
  6. Commit + checkpoint. One commit for the task, then update docs/state/CURRENT.md (done, next action, dead ends, uncommitted state) — that's the one required write. If this plan keeps a *-progress.md ledger, tick its checkbox too; add a session-log line only when you stop, and an .agents/memory/ note only when you learned something durable. Don't let the bookkeeping outweigh the task.

Whole-branch review before FINAL

After the last task, review the entire branch diff, not just the per-task diffs — cross-task seams hide bugs no single task's review could catch. Only then run the plan's Final task (fold into docs/system/, reconcile README/docs/skills, set Status: FINAL, archive) per checkpointing-progress.

Conventions

  • One commit per task; small reviewable diffs beat one big drop. Don't push unless asked.
  • The checkpoint in step 6 is what makes the work resumable by a different provider — it is not optional convenience. Update CURRENT.md regardless of which tool you're in, or the next provider loses the thread.

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.