Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsCommunityBlog
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

Back to skills

Agent Impl Notes Log

ASecurity

Keep a running impl-notes file next to the phase meeting log while a subagent executes a task, recording design decisions, intentional deviations from spec, tradeoffs, and open questions as they happen rather than after the fact. Use at the start of any non-trivial subagent dispatch (multi-file change, ambiguous spec, new dependency or module, refactor), when ambiguity is hit mid-task, or before deviating from spec. Not the post-hoc meeting log (session-to-meeting-log); the two coexist.

18 stars
0 votes
0 copies
0 views
Added 9/19/2026
ai-agentsgodocumentation

Security Analysis

A100/100

Scanned 9/19/2026

Install to Claude Code

$npx -y skills add wei18/apple-dev-skills --skill agent-impl-notes-log --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Agent Impl Notes Log?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Agent Impl Notes Log
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/wei18-agent-impl-notes-log/badge)](https://www.skillsdirectory.com/skills/wei18-agent-impl-notes-log)

More formats (shields.io, HTML) on the badges page.

Download Zip
Files
SKILL.md
---
name: agent-impl-notes-log
description: "Keep a running impl-notes file next to the phase meeting log while a subagent executes a task, recording design decisions, intentional deviations from spec, tradeoffs, and open questions as they happen rather than after the fact. Use at the start of any non-trivial subagent dispatch (multi-file change, ambiguous spec, new dependency or module, refactor), when ambiguity is hit mid-task, or before deviating from spec. Not the post-hoc meeting log (session-to-meeting-log); the two coexist."
argument-hint: "[topic]"
---

# Agent Implementation Notes — Running Log

Invoke with a topic argument (frontmatter [`argument-hint: "[topic]"`](https://code.claude.com/docs/en/skills#pass-arguments-to-skills)).

## Purpose

The phase meeting log (`meetings/{date}_{topic}.md`) is summative — written after work completes. By that point, dozens of micro-decisions made mid-flight are already lost to the commit diff. An impl-notes log fills that gap: a **concurrent record** of decisions, deviations, tradeoffs, and unresolved questions, written as they happen.

## When to invoke

Subagent MUST invoke this skill at the start of any dispatch matching ANY of:

- Implements behavior whose spec has known ambiguity (e.g., `// UNCONFIRMED` markers, "Unconfirmed ?" prerequisites).
- Introduces a new dependency, target, or module.
- Refactors existing code beyond a one-line fix.
- Any M- or L-size task by your workflow's task-sizing convention, if it has one (touches ≥2 files, adds new behavior, or gets a spec/plan before code).

Subagent MAY skip this skill for trivial one-line fixes, pure typo corrections, or documentation copy edits.

## File layout

Path: `meetings/{YYYY-MM-DD}_{topic}.impl-notes.md`

Paired naming with the phase meeting log. If the phase log is `meetings/2026-05-20_phase-11-foo.md`, the impl-notes is `meetings/2026-05-20_phase-11-foo.impl-notes.md`.

If multiple subagents work on the same topic on the same day, append a numeric suffix: `meetings/2026-05-20_topic.impl-notes.2.md`.

## File format

```markdown
# Impl Notes — {topic} ({date})

Status: IN_PROGRESS
Owner: {subagent type, e.g. "Senior Developer"}
Dispatched by: Leader
Started: {ISO timestamp}

## 設計決定 (Design decisions)

_What you chose when the spec was ambiguous. Include the spec reference._

- **{decision name}** — Spec `docs/design.md §X.Y` says "...". Ambiguous on Z. Chose **A** because [reason]. Alternative was B (see §折衷).

## 偏離 (Deviations)

_Intentional deviations from the documented spec. Include the spec reference and the why._

- **{deviation name}** — Spec says X. Implemented as Y because [reason]. Downstream impact: [what changes for callers / tests / docs].

## 折衷 (Tradeoffs)

_Alternatives considered and the reasoning for the current pick._

- **{tradeoff topic}** — Considered: [A, B, C]. Picked **B** because [reason]. Rejected A because [reason]; rejected C because [reason].

## 未決 (Open questions)

_Things you want Leader / User to confirm. Be specific. Block on these before final report if they're load-bearing._

- **{question}** — [specific question]. Default behavior I picked: [X]. Risk if I'm wrong: [impact].
```

## How to update

- **Update incrementally** — write the entry as you make the decision, NOT in a batch at the end. The whole point is to capture context that fades.
- **Keep entries terse but self-contained** — one bullet should be readable months later without re-reading the diff.
- **Cite spec sections** by section number (`§How.4.7`) so Leader can verify against the source.
- **Cross-link related entries** if a 折衷 decision later becomes a 偏離.
- **Don't duplicate the diff** — the file is for context, not code listings.

## Status transitions

- `IN_PROGRESS` — work ongoing; file may be edited any time.
- `COMPLETE` — work finished, ready for Leader review. Set this before reporting back.
- `BLOCKED` — work paused awaiting Leader/User answer to an §未決 item. Required when an open question is load-bearing.

## Subagent dispatch contract (Leader-side)

When Leader dispatches a subagent matching the invoke criteria, the dispatch prompt MUST include:

> Create `meetings/{date}_{topic}.impl-notes.md` at the start of your work using the `collaboration-skills:agent-impl-notes-log` skill format (or have it preloaded via the agent definition's [`skills:` field](https://code.claude.com/docs/en/sub-agents#preload-skills-into-subagents)). Update it incrementally — design decisions, deviations, tradeoffs, open questions. Mark `Status: COMPLETE` before final report. Include the file path in your report.

Subagent's final report MUST link the impl-notes file. Leader reads it before merging the work to verify decisions match Leader's intent.

## Anti-patterns

- **Writing impl-notes at the end as a batch** — defeats the purpose; you've already forgotten the context.
- **Logging every minor edit** — file is for *decisions*, not edit history. Trivial mechanical fixes belong in the commit, not here.
- **Using this file instead of the phase meeting log** — the two coexist. Phase log is summative, impl-notes is concurrent. Phase log gets archived when sealed; impl-notes can be referenced from the phase log.
- **Sealing the file without resolving 未決 items** — open questions must be answered (by Leader/User) or moved to a backlog before status flips to COMPLETE.

## Related skills

- `session-to-meeting-log`: produces the post-hoc phase meeting log; the two artifacts pair.
- `leader-developer-handoff-contract`: defines the dispatch contract; impl-notes is one of its outputs.
- `subagent-review-cycles`: when a subagent's work returns for review, Leader reviews the impl-notes alongside the diff.

Attribution

wei18wei18
View sourceMore from wei18 →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accuracy. Supports intensity levels: lite, full (default), ultra, wenyan-lite, wenyan-full, wenyan-ultra. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman. Also auto-triggers when token efficiency is requested.

1023331 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

686011 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3331 votes

catchup

Recovers prior coding-agent session context by running `catchup <agent> --since-compact`, which extracts a clean summary of a previous Codex, Claude Code, Antigravity, OpenCode, or Pi Agent session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", or asks to recover/summarize a previous session before continuing. Do NOT use for the current conversation, git history, or any non-agent log.

611 votes

math-skill

A comprehensive mathematical reasoning skill for AI assistants — handles arithmetic to research-level problems with rigorous step-by-step reasoning, systematic verification, and transparent uncertainty handling

381 votes
View all in ai-agents →