agentleFS
forsonnyforsonny/DG-Anvil/skills/planning/SKILL.md

planning

Skill0 starsChanged 7 months agoMITView on GitHub

Decompose a confirmed contract into an atomic task DAG with wave ordering. Every task cites one or more contract criterion ids.

69 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use Problem-Solving Techniques, web-frameworks and 8 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.
  • truongtv22Skill

    Problem-Solving Techniques

    Apply systematic problem-solving techniques for complexity spirals (simplification cascades), innovation blocks (collision-zone thinking), recurring patterns (meta-pattern recognition), assumption constraints (inversion exercise), scale uncertainty (scale game), and dispatch when stuck. Techniques derived from Microsoft Amplifier project patterns adapted for immediate application.

    Also in 5 projects that use this

  • mrgoonieSkill

    web-frameworks

    Build modern full-stack web applications with Next.js (App Router, Server Components, RSC, PPR, SSR, SSG, ISR), Turborepo (monorepo management, task pipelines, remote caching, parallel execution), and RemixIcon (3100+ SVG icons in outlined/filled styles). Use when creating React applications, implementing server-side rendering, setting up monorepos with multiple packages, optimizing build performance and caching strategies, adding icon libraries, managing shared dependencies, or working with TypeScript full-stack projects.

    Also in 5 projects that use this

  • mrgoonieSkill

    mcp-management

    Manage Model Context Protocol (MCP) servers - discover, analyze, and execute tools/prompts/resources from configured MCP servers. Use when working with MCP integrations, need to discover available MCP capabilities, filter MCP tools for specific tasks, execute MCP tools programmatically, access MCP prompts/resources, or implement MCP client functionality. Supports intelligent tool selection, multi-server management, and context-efficient capability discovery.

    Also in 5 projects that use this

  • truongtv22Skill

    mobile-development

    Build modern mobile applications with React Native, Flutter, Swift/SwiftUI, and Kotlin/Jetpack Compose. Covers mobile-first design principles, performance optimization (battery, memory, network), offline-first architecture, platform-specific guidelines (iOS HIG, Material Design), testing strategies, security best practices, accessibility, app store deployment, and mobile development mindset. Use when building mobile apps, implementing mobile UX patterns, optimizing for mobile constraints, or making native vs cross-platform decisions.

    Also in 4 projects that use this

  • +6 more from mrgoonie

The file

We split this skill into the rules it gives an agent

5 rules. 0 are common, 2 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 62 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

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

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

    Read the returned one-sentence summary from the subagent.

    line 22
  2. Distinctive:

    If any of these conditions obtain, the plan is rejected:

    line 36
  3. Distinctive:

    Before saving the plan, run in order:

    line 45
  4. Very specific:

    Every contract criterion id appears in at least one task's criterion_ids array.

    line 48
  5. Very specific:

    Every task in wave k > 0 has a depends_on whose members are tasks in waves < k.

    line 50

“Common” is measured: kept by at least 6 in 10 of the 62 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

Overview

This skill produces anvil/plan.yml as a task DAG with wave ordering. An atomic task is one that produces one diff that one Verify pass can score. Every task names the contract criterion id(s) it is accountable for; tasks without citations are invalid and do not execute. Waves are topologically sorted: a task in wave k depends only on tasks in waves strictly less than k.

Invoking the Anvil CLI: the CLI is cli/anvil.js inside the plugin directory, NOT on PATH. Use node "$CLAUDE_PLUGIN_ROOT/cli/anvil.js" <subcommand> ... for every invocation.

When to Use

Invoked by the orchestrator after anvil/contract.yml is confirmed. Not invoked from user prompts directly. If the contract is edited after the plan is written, this skill is re-invoked to regenerate the plan from scratch.

Process

Steps 1 and 4 run in the orchestrator's main thread. Steps 2 and 3 (decomposition + YAML authoring) are dispatched to a fresh subagent. The orchestrator holds only the confirmed contract and the written anvil/plan.yml file. It never holds the decomposition reasoning.

  1. Confirm anvil/contract.yml exists and validates (node "$CLAUDE_PLUGIN_ROOT/cli/anvil.js" contract --validate anvil/contract.yml exits 0). If it does not, route back to contracting; planning does not run against an unconfirmed contract.
  2. Dispatch the plan-drafter agent (defined at agents/plan-drafter.md) using the Task tool. Pass in the briefing: the repository path and the path to anvil/contract.yml. The subagent reads the contract, decomposes the work into an atomic task DAG with wave ordering, writes anvil/plan.yml, runs anvil plan --validate, and returns exactly one sentence. Do NOT decompose or write the YAML inline in the orchestrator thread - the orchestrator's context stays clean.
  3. Read the returned one-sentence summary from the subagent. If the summary indicates a validation failure, surface the details.rule string to the user and stop.
  4. The orchestrator now holds the task DAG by file reference (anvil/plan.yml). It does not re-read the whole plan for every subsequent dispatch; it reads individual task records as the executing skill requests them.

Rationalizations

Reject the following shortcuts:

  • "This one task covers everything; no need to split." Scope creep disguises itself as efficiency; an atomic task is scorable, a combined task is not (failure-taxonomy row 4: Scope creep).
  • "Over-produce now and trim later." Over-production burns context and muddies the Verify signal (failure-taxonomy row 10: Over-production).
  • "The plan can drift from the spec; the agent will catch it later." Plan drift is silent until Verify fails on a criterion no task cited (failure-taxonomy row 26: Spec-to-plan drift).
  • "I'll decompose the tasks inline; it's faster." Inline decomposition fills the orchestrator context with reasoning that should live in a fresh subagent (failure-taxonomy row 16: Context-window collapse).

Red Flags

If any of these conditions obtain, the plan is rejected:

  • A contract criterion id does not appear in any task's criterion_ids (failure-taxonomy row 26: Spec-to-plan drift).
  • A task's scope extends beyond the criterion it cites (failure-taxonomy row 4: Scope creep).
  • Two tasks in different waves touch the same file without a declared depends_on edge (failure-taxonomy row 17: Cross-task architectural drift).
  • The orchestrator decomposes tasks in its own thread instead of dispatching the plan-drafter agent (failure-taxonomy row 16: Context-window collapse).

Verification

Before saving the plan, run in order:

  1. node "$CLAUDE_PLUGIN_ROOT/cli/anvil.js" plan --validate anvil/plan.yml --contract anvil/contract.yml exits 0.
  2. Every contract criterion id appears in at least one task's criterion_ids array.
  3. The task DAG, interpreted via topologicalWaves, produces one wave per distinct wave integer and contains no cycles.
  4. Every task in wave k > 0 has a depends_on whose members are tasks in waves < k.

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.