Convert Eip

Convert EIP to forkcast format

Convert an EIP proposed for the current fork (e.g., Glamsterdam, Hegota) into forkcast JSON format.

Scope note: this is the whole-file workflow. If the EIP already exists in src/data/eips/ and only needs its laymanDescription, benefits, and tradeoffs filled in, use the draft-eip-narrative skill instead — it is the reduced-scope version of the same field rules.

Schema

Get the current schema:

• Type definitions: src/types/eip.ts
• Example EIPs: src/data/eips/*.json

Writing Guidelines

• description: max 80 words, match the EIP abstract as closely as possible
• laymanDescription: max 60 words, plain language for non-technical readers
• layer: choose EL or CL, whichever is more appropriate
• reviewer: set to "bot" for AI-generated conversions like this
• stakeholderImpacts: ~20 words each
  • clClients and elClients: focus on implementation complexity
• benefits: max 16 words each, up to 4
• tradeoffs: max 16 words each, include if any exist
• discussionLink: use the EIP's discussions-to URL from the frontmatter. If empty (e.g., unmerged EIP), use the headliner proposal URL or Eth Magicians thread if available.
• specificationUrl: only set this for unmerged EIPs where the default eips.ethereum.org URL would 404. Point it to the GitHub PR (e.g., https://github.com/ethereum/EIPs/pull/11376).
• northStarAlignment: include if EIP aligns with any of these goals (1 sentence each):
  • scaleL1: L1 throughput/efficiency improvements
  • scaleBlobs: Blob capacity/scaling improvements
  • improveUX: User or developer experience improvements

For descriptions, impacts, and benefits: be as factual and true to the resources as possible. Do not speculate. Do not shoehorn impacts if none exist. Do not assume any information. Return output in code format, without citations.

Step 1: Gather inputs

Ask user to fill in and paste:

eip: 7807
call: acde/229
discord: @handle
headliner: https://ethereum-magicians.org/... (or "no")
status: (optional, e.g. "Proposed" or "Considered" - leave blank if none)
context: (optional eth r&d discord context)

Status & Presentation History

statusHistory: Add if user provides status or transcript has explicit status change. Don't infer—leave empty if uncertain.

presentationHistory types:

• headliner_proposal: Eth Magicians post proposing EIP as headliner. Requires link field.
• headliner_presentation: First presentation at a call as headliner candidate (use when user provided a headliner URL). Requires call and date fields.
• presentation: General presentation at a call for non-headliner EIPs. Requires call and date fields.
• debate: Follow-up discussion at a later call, after initial presentation. Requires call and date fields.

Do not add timestamp fields unless you are sure they refer accurately to the start time.

Headliner Flags

If the user provides a headliner URL (i.e., headliner is not "no"), set:

• isHeadliner: false
• wasHeadlinerCandidate: true
• For the call presentation, use headliner_presentation (not debate or presentation)

Step 2: Fetch resources

Use the EIP number to gather all resources:

2a. Raw EIP (Latest from Master)

Fetch the current version from master:

gh api '/repos/ethereum/EIPs/contents/EIPS/eip-{EIP_NUMBER}.md' --jq '.content' | base64 -d

If this 404s (unmerged EIP), fall back to fetching from the PR branch instead. Find the PR, get the head SHA, and fetch from that ref.

2b. Commit History

Get all commits that modified this EIP:

gh api '/repos/ethereum/EIPs/commits?path=EIPS/eip-{EIP_NUMBER}.md' --jq '.[] | {sha: .sha[0:7], date: .commit.author.date[0:10], message: .commit.message | split("\n")[0]}'

Review these to understand how the EIP evolved. The original "Add EIP" commit is typically the last/oldest one.

2c. Original PR Discussion

Find the original PR from the first commit:

# Get the original commit SHA (last in list = oldest). Validate it looks like a hex SHA before using.
ORIGINAL_SHA=$(gh api '/repos/ethereum/EIPs/commits?path=EIPS/eip-{EIP_NUMBER}.md' --jq '.[-1].sha')

# Find the PR for that commit
gh api "/repos/ethereum/EIPs/commits/$ORIGINAL_SHA/pulls" --jq '.[0] | {number, title, html_url}'

Then fetch the full PR discussion. Important: Must fetch BOTH issue comments AND review comments (line-level):

# Issue-level comments
gh pr view {PR_NUMBER} --repo ethereum/EIPs --json title,body,author,createdAt,comments

# Review comments (line-level, includes resolved comments)
gh api /repos/ethereum/EIPs/pulls/{PR_NUMBER}/comments --jq '.[] | {user: .user.login, body: .body, created_at: .created_at}'

# Reviews (approvals, change requests)
gh api /repos/ethereum/EIPs/pulls/{PR_NUMBER}/reviews --jq '.[] | {user: .user.login, state: .state, body: .body}'

Important: Verify you found the original "Add EIP" PR (title should start with "Add EIP"). If not found, ask the user for the original PR link.

2d. Eth Magicians Discussion

Do NOT use WebFetch - it summarizes and loses detail. Use the JSON API directly:

# Get thread metadata and post count
curl -s "https://ethereum-magicians.org/t/{TOPIC_SLUG}/{TOPIC_ID}.json" | jq '{title, posts_count, created_at}'

# Get all posts with content
curl -s "https://ethereum-magicians.org/t/{TOPIC_SLUG}/{TOPIC_ID}.json" | jq '.post_stream.posts[] | {username, created_at, cooked}'

Extract the topic slug and ID from the discussions-to URL (e.g., eip-7807-ssz-execution-blocks/21580).

Note: Discourse only returns the first 20 posts by default. For threads with many posts, fetch the latest posts by using the post IDs from post_stream.stream (which contains all IDs) and requesting the most recent ones via ?post_ids[]=.

2e. Headliner Proposal

If user provides a headliner URL, fetch it with the same JSON API approach:

curl -s "https://ethereum-magicians.org/t/{TOPIC_SLUG}/{TOPIC_ID}.json" | jq '.post_stream.posts[] | {username, created_at, cooked}'

Extract the post date from created_at. Do not assume it matches the call date.

2f. Call Transcript

Glob: public/artifacts/{call_type}/*{number}*/**. Prefer transcript_corrected.vtt, fall back to transcript.vtt. Grep for the EIP number or title to find relevant discussion.

2g. Related EIPs

If the EIP has a requires field, fetch those using the same method.

2h. Champion Name

Extract from the EIP author field (the name before the GitHub handle). Discord handle comes from user input.

Step 3: Pre-generation check

Before generating JSON, verify you have enough information for all required fields. If missing critical info (e.g., can't determine layer, no clear benefits from sources), ask user before proceeding.

Step 4: Generate output

Generate files in this order:

1. src/data/eips/{EIP_NUMBER}-context.md - context file first (raw data, local reference only, not committed)
2. src/data/eips/{EIP_NUMBER}.json - EIP JSON second (synthesized from context)

Context File Format

The context file preserves the FULL raw data used to generate the EIP. Every section must include its source URL or file path so readers can trace where each excerpt came from.

# EIP-{number} Context

Generated: {date}

## Raw EIP Content
Source: https://github.com/ethereum/EIPs/blob/master/EIPS/eip-{number}.md
\`\`\`
{full raw EIP markdown}
\`\`\`

## Commit History
Source: https://github.com/ethereum/EIPs/commits/master/EIPS/eip-{number}.md
\`\`\`
{full commit history output}
\`\`\`

## Original PR Discussion
Source: {pr_url}

### PR Body
{full pr body}

### Issue Comments
{all issue-level comments with author/date}

### Review Comments
{all line-level review comments with author/date}

## Eth Magicians Discussion Thread
Source: {thread_url}

### Posts
{all posts with username/date/content}

## Headliner Proposal (if applicable)
Source: {headliner_url}

### Posts
{all posts with username/date/content}

## Call Transcript
Source: {call_ref} - {transcript_path}

### Relevant Excerpts
{excerpts with timestamps}

## Eth R&D Discord Context (if provided)
Source: Eth R&D Discord (user-provided)
{user-provided discord context}

## Related EIPs (if any)
{summaries of required EIPs with links}

Step 5: Validate

Run the compile script to verify the generated EIP is valid:

npm run compile-eips

Then run the metadata validator to auto-fix fields like title, description, status, and author from the canonical EIP source:

npm run validate-eips -- --eip {EIP_NUMBER} --fix

If there are errors in either step, fix them before reporting success.

Step 6: PR

Important: Only commit the .json file. Do not commit the *-context.md file—it is for local reference only.

PR Title: Add EIP-{number}: {title}

PR Body:

> [!NOTE]
> This PR was generated with the `convert-eip` skill (see [SKILL.md](https://github.com/ethereum/forkcast/blob/main/.agents/skills/convert-eip/SKILL.md)).

## EIP-{number}: {title}

Ported to forkcast format.

### Files
- `src/data/eips/{number}.json` - EIP data

### Sources Used
| Source | Reference |
|--------|-----------|
| Raw EIP | ethereum/EIPs/EIPS/eip-{number}.md |
| Original PR | #{pr_number} - {pr_title} |
| Commits | {count} commits ({date range}) |
| PR Discussion | {comment_count} comments |
| Eth Magicians | {thread_title} ({post_count} posts) |
| Call Transcript | {call_ref} |
| Eth R&D Discord | {yes/no} |
| Headliner Proposal | {url or "N/A"} |

### Metadata
- **Champion**: {name} ({discord_handle})
- **Layer**: {EL/CL}
- **Status**: {status or "none"}
- **Headliner**: {yes/no}

Step 7: Retrospective

After completing the PR, review the run for any friction: missing guidance, incorrect defaults, assumptions that needed correction, or steps that broke. Present a hyphenated list of issues to the user and offer to patch this skill to address them.