Oxyflax AI Copilot Migration

Copilot → Claude Code migration

Bring a user's GitHub Copilot conversation history into Claude Code in three steps: extract the raw conversations from disk → document them as readable Markdown plus an index → continue the most recent one.

GitHub Copilot stores chat history locally in VS Code's storage (two different on-disk formats, see the reference). Nothing is fetched from the network.

When to use

• "extract / recover / export / back up my Copilot conversations or chat history"
• "I'm migrating from GitHub Copilot to Claude Code"
• "document / summarize what I did with Copilot"
• "continue my last Copilot conversation" / "pick up where Copilot left off"

What's bundled (resolve paths from this skill's directory)

• scripts/extract_copilot.py — standalone, no-dependency extractor. Discovers Copilot storage across editors/OSes, decodes both on-disk formats, dedupes sessions resumed across windows, and writes one faithful Markdown transcript per conversation + a manifest.json. Run it — do not re-implement the parsing.
• references/storage-and-formats.md — exactly where Copilot stores chats per OS/editor and how the two formats work. Read it if discovery finds nothing, or to explain the mechanics to the user.

Procedure

1 — Extract (deterministic, read-only)

List first, then extract:

python3 <skill>/scripts/extract_copilot.py --list           # discover + preview
python3 <skill>/scripts/extract_copilot.py --out <out-dir>  # write transcripts

• Pick <out-dir>: if the project has a docs/ directory, use docs/copilot-history/; otherwise ./copilot-history/.
• To scope to one project, add --filter <substring> (matches the workspace path or conversation content). Use --no-reasoning to drop the model's "thinking".
• If --list finds nothing: read references/storage-and-formats.md, check the candidate paths it printed to stderr, and ask the user which editor/host they used Copilot in (desktop vs. remote SSH/WSL/Cursor change the location).
• Output: <out-dir>/transcripts/*.md (faithful, verbatim) + manifest.json. Skim manifest.json (title/date/turns/files per conversation) to decide what matters — don't blindly re-read every large transcript.

2 — Document (you write these)

In <out-dir>:

• sessions/<same-filename>.md — a tight summary per conversation: title, date, goal, what was done, files changed, key decisions, outcome, open threads. Read the full transcript for large/important ones; the manifest + first/last turns suffice for trivial ones. If there are many, fan out subagents (one per transcript).
• README.md — an index: a 2–3 sentence intro, a chronological table (date | linked conversation | one-line outcome), and a grouping by theme.

3 — Continue the most recent conversation

This is usually the user's real goal.

• The latest conversation is the last row of manifest.json — it is sorted by most-recent activity (last_ts), so the final row is the conversation the user used most recently (each row also carries first_ts if you want start-date order).
• Read its full transcript. Cross-reference against the current repo state (git log, the files it touched) to separate what actually landed from what was only discussed.
• Write <out-dir>/CONTINUE_HERE.md: where it left off (exact stop point), what landed vs. discussed, verified vs. unverified, open problems, and a prioritized next-steps checklist referencing concrete files/commands.
• Then offer to do the next step.

Notes & guarantees

• The extractor is read-only on Copilot storage and writes only under --out. No network, no extra packages (standard library only). Safe to run anywhere.
• A conversation resumed across several editor windows is saved as multiple files sharing one sessionId (later windows replay earlier turns); the script merges and dedupes turns by prompt content, keeping the most complete copy.
• The division of labor is deliberate: the script does the exact, verbatim extraction (no hallucination risk); you do the summarizing and judgement.
• This skill is distributable: it has no hard-coded paths or project assumptions — discovery and --filter are fully general.

Source: OxyFlax/AI — distributed by TomeVault.