SkillAgentSearch skills...

implement-plan

Executes a complete, approved engineering plan through exactly one implementation agent. Use after planning is finished and material decisions are settled.

Install / Use

npx skills add rphlmr/claude-config --skill implement-plan

Installs into whichever agent you are using.

About this skill
📄

SKILL.md

Installable skill definition

Quality Score

62/100

Supported Platforms

Universal

Tags

Our assessment of implement-plan

implement-plan scores 62/100 on our quality scale, 2064th of 2,717 Development & Engineering skills we index.

Its SKILL.md is 6.5 KB long, well organised into 9 sections and no code examples: a thorough specification that gives an agent plenty to work with.

It has no GitHub stars yet, so there is no community track record; judge it on its content.

Substance
29/30
Structure
13/20
Description
15/15
Adoption
0/20
Freshness
5/15

Maintenance, license and trust

  • We could not determine when the repository was last updated.
  • No license is declared. By default that means all rights are reserved: you can read it, but reusing or redistributing it is not clearly permitted. Ask the author before building on it commercially.
  • Its trust signals score 68/100, with 3 cautions from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

Safety scan

No issues found

Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.

AI review by kimi-k2.7-code on 2026-09-27. Automated pattern scan on 2026-09-27. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.

implement-plan compared with similar skills

All 4 of these similar skills score higher than implement-plan; compare them before choosing.

SkillScoreStarsUpdatedFormat
implement-plan (this skill)by rphlmr620—SKILL.md
ai-job-searchby MadsLorentzen10044.1ktodayCLAUDE.md
claude-howtoby luongnv8910041.7k1d agoCLAUDE.md
algorithmic-artby anthropics100177.9k5d agoSKILL.md
pptxby anthropics100177.9k5d agoSKILL.md

Frequently asked questions

How do I install implement-plan?
Run npx skills add rphlmr/claude-config --skill implement-plan. The install tabs above show the steps for each supported agent.
Which AI agents does implement-plan work with?
It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
Is implement-plan safe to use?
Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It declares no license and scores 68/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is implement-plan still maintained?
We could not determine when the repository was last updated.

name: implement-plan description: Executes a complete, approved engineering plan through exactly one implementation agent. Use after planning is finished and material decisions are settled. disable-model-invocation: true

Implement Plan

Execute the current approved plan through exactly one implementation subagent.

This is an execution handoff. Do not perform another planning, investigation, review, or implementation phase in the parent thread.

Establish the implementation brief

Treat the approved plan as current.

Pasted plan

When the invocation arguments contain a complete implementation plan:

  • use that plan verbatim;
  • do not summarize, restructure, normalize, or reinterpret it;
  • include only explicit amendments stated after the plan.

Plan from the current conversation

Otherwise:

  • use the latest explicitly approved final plan from the current conversation;
  • preserve its wording and structure;
  • append only later explicit decisions or corrections;
  • exclude exploratory discussion, rejected alternatives, superseded plans, and unrelated conversation.

Do not inspect the repository to verify plan freshness.

Do not recreate an already complete plan as a shorter brief. Rewriting can omit important compatibility, validation, or acceptance details.

If no complete approved plan can be identified, stop. Do not create a new plan under this skill.

Treat a plan as complete only when it contains, at the level required by the task:

  • a clear outcome;
  • executable implementation boundaries or steps;
  • settled material decisions and constraints;
  • objectively checkable acceptance criteria;
  • required validation commands or validation expectations.

For a simple change, these elements may be concise or combined. Do not require ceremonial sections when the required information is already explicit.

If a material element is missing, stop and identify the missing element. Do not infer, reconstruct, or repair the plan under this skill.

Validation contract

Treat the validation specified by the approved plan as the complete required validation set.

  • Preserve exact validation commands when they are present in the approved plan or were explicitly added as a later approved amendment.
  • Do not add broader test suites or validation categories.
  • Do not remove or weaken required validation.
  • Do not convert descriptive validation requirements into guessed commands in the parent thread.

The implementation agent may resolve exact repository commands when they were not established during planning.

Select the implementation agent

Select the implementation agent from the approved plan before spawning it. light-implementer is the lighter executor for narrow mechanical work.

Use implementer when the plan changes any of:

  • exported or public TypeScript APIs;
  • conditional, recursive, nominal, or inference-heavy types;
  • emitted declarations;
  • shared monorepo contracts;
  • compatibility or migration surfaces;
  • authentication, security, persistence, or build infrastructure.

Use light-implementer for:

  • localized runtime behavior;
  • mechanical refactors with exact instructions;
  • narrow UI changes;
  • documentation;
  • repetitive test additions;
  • corrections where a verifier has already identified the exact symbols and expected types.

The implementer criteria take precedence when categories overlap. Honor an explicit user request for either implementation agent.

Spawn exactly one selected implementation agent with the Agent tool, as a new subagent of that type rather than a fork of this conversation:

  • implementer for the contract-heavy and high-risk categories above;
  • light-implementer for the narrow and mechanical categories above.

Do not spawn additional implementation, exploration, planning, review, or verification agents unless the user explicitly requests a separate review.

Handoff

The agent starts without this conversation, so the handoff is its entire brief. Send the selected agent this short execution contract:

The following implementation plan is current, approved, and authoritative. Implement it directly under your standing agent instructions.

Treat its validation section as the complete required validation set.

Append the approved implementation plan verbatim.

Do not duplicate the implementation agent's standing instructions in the handoff.

Wait for the selected agent to return. Its result may arrive later as a completion notification; do not report an outcome before it arrives.

Do not inspect, implement, validate, or review the same changes independently in the parent thread.

Material decision escalation

When the implementation agent returns BLOCKED_DECISION, the parent owns the decision.

Use the reported repository evidence together with:

  • the approved plan;
  • explicit decisions already made;
  • active instructions;
  • the requested outcome.

When those establish one unambiguous answer:

  1. make the decision in the parent thread;
  2. send only the resolved decision to the same implementation agent with SendMessage, addressed by the agent ID or name from its result;
  3. instruct that agent to continue.

Ask the user only when multiple valid outcomes still depend on product intent or preference.

Do not:

  • ask the implementation agent to choose an architecture;
  • ask it to recommend between product or API alternatives;
  • replace it because it escalated;
  • spawn a second implementation agent;
  • send the complete plan again unless the agent requests missing context.

Completion

When the implementation agent completes:

  • inspect its completion report and require:
    • one line for every acceptance criterion;
    • the concrete test, declaration, or behavior proving each criterion;
    • every exact required validation command and its result;
    • every declaration-inspection requirement and its result;
    • explicit Unresolved and Unverified sections, using None when empty;
  • if any required report element is absent, send the same implementation agent a targeted follow-up with SendMessage identifying the missing report elements, and wait for its corrected completion report before proceeding;
  • report the implementation outcome;
  • include acceptance-criterion results;
  • include validation commands and results;
  • include deviations, unverified items, and unresolved issues;
  • clearly state any failed required validation.

Do not redo the implementation or validation in the parent thread.

Do not claim successful completion when a required acceptance criterion remains unsatisfied or required validation failed.

Related Skills

View on GitHub
GitHub Stars0
CategoryDevelopment
UpdatedNaNy ago
Forks0

Trust signals

68/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

2 medium1 low