메뉴
Surface recurring friction from Claude Code's native telemetry and identify actionable skill opportunities. Use when the user runs /forge-friction.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
Release pipeline for already-done local work. Dispatches /ground-state pre-flight, runs the project test suite, drafts a commit message for user approval, pushes, and opens a PR with a structured verification summary. Use when local changes are ready to hand off to review — e.g. 'ship this', 'push and open a PR', 'release this work'. Add --verify to trigger an adversarial verifier wave on the diff before a human reads the PR.
Bootstraps a dispatchable Claude sub-agent skill for every coding agent CLI installed on this machine (codex, aider, cursor, etc.). Use when the user wants to delegate tasks to locally-installed agent CLIs or expand the plugin's agent fleet.
Map a web UI and generate a Claude automation skill via parallel reconnaissance, synthesis, and iterative validation.
Set up a scheduled headless Claude run that sends summaries to Telegram. Use when the user wants to automate a recurring task via launchd (macOS) with push-notified results.
Reference convention for sub-agent I/O schemas. Loaded by orchestrator skills via /agent-workflow-amplifiers:contract and into agents (e.g., qualify) via the `skills:` field.
Before starting any non-trivial implementation (multi-file edits, new features, config changes, anything that writes), dispatch a parallel pre-flight reconnaissance wave to triangulate git state, project infrastructure, and prior-session memory. Produces a 5-line ground-truth snapshot that grounds the implementation and catches wrong-branch edits, assumed-no-CI, stale origin, and missed memory context before the first edit.
SOC 직업 분류 기준
| name | forge-friction |
| description | Surface recurring friction from Claude Code's native telemetry and identify actionable skill opportunities. Use when the user runs /forge-friction. |
Run the friction analyzer to get a summary of recent friction patterns:
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/friction/analyzer.py"
If no friction sessions exist, tell the user there are no friction patterns in the lookback window.
If friction data exists, review the output. It contains friction categories ranked by frequency, each with recent examples showing friction_detail (what went wrong) and goal (what the user was trying to do).
Read through the friction categories and their examples. For each category with 3+ sessions, analyze the friction_detail strings to identify recurring themes — specific, repeated failure modes that a skill could address.
Not all friction is fixable by a skill. Filter for themes where:
Present your findings to the user:
For each actionable theme: summarize it in one line with the count, then quote 2-3 representative friction_detail examples. Ask: "Draft a skill brief for this? (y/n)"
If the user says yes: Output a structured skill brief for the theme (NOT an actual skill — that's a separate workflow) — a one-paragraph description of what the skill should do, the friction it addresses, and 2-3 representative examples. Also persist the brief to ~/.claude/agent-framework/briefs/<ISO8601-compact>-<theme-slug>.md (e.g., 2026-04-17T14-30-05-wrong-approach-refactor.md), creating the directory if needed. Use this format:
---
theme: <one-line theme>
session_count: <int>
created_at: <ISO8601>
source: forge-friction
---
<brief paragraph>
## Friction examples
- <example 1>
- <example 2>
- <example 3>
The user can feed the brief into their own skill creation workflow. Downstream tools in separate plugins (e.g., autonomous skill generators) can also consume the persisted briefs.
If the user says no, skip it and move on.
You can also drill into a specific category:
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/friction/analyzer.py" --category wrong_approach
Or adjust the lookback window:
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/friction/analyzer.py" --days 60
After processing, summarize: how many themes found, how many the user approved, how many skills generated.